Patent Application
20080303943
This application claims priority under 35 U.S.C. §119(a) on Japanese Patent Application No. 2007-149703 filed on Jun. 5, 2007, the entire contents of which are hereby incorporated by reference.
1. Field of the Invention
The present invention relates to recording and playback of a television broadcast program, and more particularly relates to a technique for generating a digest of a television broadcast program.
2. Description of the Related Art
Some of hard disk recorders for recoding and playing back a television broadcast program have the function of playing back a digest of the program. In such systems, digest scenes are extracted and played back according to a genre designated by a user. Genres for which digests are generated include music programs, sports programs such as baseball, soccer, golf, etc., and the like. For example, in the case of a baseball game, scenes of homeruns and strikeouts are scenes which an audience wants to watch as a digest and, in the case of a soccer game, scenes of plays leading to a goal are such scenes.
In a known technique, scenes where a video image is largely changed and an audio level is larger than a predetermined value are detected as digest scenes (see, for example, Japanese Laid-Open Publication No. 2005-348077).
In the genres of programs such as drama, movies and the like, having a story plot, scenes where a person utters are more important in comprehending an overview of a program than scenes where a video image is largely changed or an audio level is increased. Therefore, if a digest is generated in a known manner for a program in which conversations of persons are dominant in the contents of the program, the generated digest might be one with which a story plot of the program is difficult to comprehend.
In view of the above-described points, it is an object of the present invention to achieve the generation of a digest of a television broadcast program of drama, movie or like genre, which allows easy comprehension of a story plot of the program.
To achieve the above-described object, means devised according to the present invention includes, as a device for generating a digest of a television broadcast program containing subtitle information, a character number calculation section for calculating, based on the subtitle information, a character number of a subtitle displayed in each of segments provided at regular intervals; and a digest scene specifying section for comparing the calculated character number with a threshold and specifying, as one or more digest scenes of the television broadcast program, one or more segments in which the calculated character number is larger than the threshold. Moreover, as a method for generating a digest of a television broadcast program containing subtitle information, means devised according to the present invention includes: a first step of calculating, based on the subtitle information, a character number of a subtitle displayed in each of segments provided at regular intervals; and a second step of comparing the calculated character number with a threshold and specifying, as one or more digest scenes of the television broadcast program, one or more of the segments in which the calculated character number is larger than the threshold.
Thus, as one or more digest scenes of a television broadcast program, one or more of segments of the television broadcast program in which the number of characters to be displayed is larger than a threshold is specified. That is, scenes containing utterance of one or more persons are extracted from the entire program and a digest of the extracted scenes is generated.
It is preferable that the character number calculation section and the digest scene specifying section calculates the character number in parallel to recording the television broadcast program and records information for the one or more specified segments in parallel to recording the television broadcast program. In the first and second steps, the same operation as the operation described above is performed. Thus, a digest can be immediately played back after recording a television broadcast program.
It is also preferable that the character number calculation section identifies speakers based on the subtitle information and calculates the character number of each of the identified speakers. Specifically, the character number calculation section obtains information for a character color of the subtitle or character information attached to the subtitle from the subtitle information and identifies speakers using the information for the character color or the character information. Then, for each of the identified speakers, the digest scene specifying section compares the calculated character number with a threshold of an associated one of the speakers and specifies, as one or more digest scenes of the television broadcast program, one or more segments in which a calculated character number of any one of the identified speakers is larger than the threshold of said any one of the speakers. In the first and second steps, the same operation as the operation described above is performed. Thus, a digest in which each of persons is weighted has been performed can be generated.
It is also preferable that the above-described digest generation device further includes a threshold setting section for summing up respective times of the one or more segments specified by the digest scene specifying section, comparing a sum of the times with a designated time and changing the threshold according to a comparison result. The digest scene specifying section respecifies the one or more segments when the threshold is changed. Thus, a digest which can fit in a designated time period can be generated.
It is also preferable that the digest scene specifying section compares the calculated character number with each of a plurality of thresholds and specifying, for each of the thresholds, one or more segments in which the calculated number is larger than an associated one of the thresholds. Then, the device further includes a digest selection section for summing up, for each of the thresholds, respective times of the one or more segments specified by the digest scene specifying section, comparing a sum of the times of each group of the segments specified for an associated one of the thresholds with a designated time and selecting, as the digest of the television broadcast program, one of the groups of the specified segments in which the sum of the times is equal to or smaller than the designated time. In the digest generation method, the same structure as the above-described structure is also used. Thus, a digest which satisfies designated time conditions can be generated in a single processing without repeating the processing of digest scene specification by changing the threshold many times.
It is also preferable that the character number calculation section reads ahead the subtitle information and calculates the character number while the television broadcast program is read out from memory means. In the first step, the same operation as the operation described above is performed. Thus, a result of specification of digest scenes of the broadcast program is not needed to be stored and therefore the size of the buffer for temporarily storing one or more specified segments as one or more digest scenes can be small.
Hereafter, best modes of the present invention will be described with reference to the accompanying drawings.
A digest generation device 60A includes a character number calculation section 62, a digest scene specifying section 64 and a threshold setting section 66. The character number calculation section 62 extracts character codes displayed as subtitles from a subtitle PES output from the TS decoder 40 and calculates a display start time and a display end time. The character codes can be extracted from subtitle data stored in a data region of the subtitle PES. The display start time can be obtained from a PTS (Presentation Time Stamp) contained in a header region of the subtitle PES. The display end time can be calculated by obtaining a display timing of a screen delete code or the like contained in the subtitle data.
Next, after extracting character codes, the character number calculation section 62 calculates the character number of a subtitle displayed for each of segments provided at regular intervals. Calculated start and end times of each segment and calculated character number are recorded in the memory means 30 or some other buffer not shown in
In some dramas and movies, subtitles are displayed such that utterances of each person distinguished from utterances of others by a different character color. Normally, such character color information is stored as a control code in the data region of the subtitle PES. Moreover, there are cases where at the beginning of subtitles, the name of a person who utters is displayed with parentheses as part of the subtitles. The character number calculation section 62 may analyze the data region to obtain character color information and parenthesized character information and use such information to distinguish a speaker and calculate the character number of each speaker.
Returning to
When the character number is calculated for each speaker by the character number calculation section 62, the digest scene specifying section 64 compares, for each speaker, the calculated character number with a threshold for the speaker given by the threshold setting section 66. Then, a segment in which the character number of any one of speakers is larger than the threshold for the speaker is specified as a digest scene of a television broadcast program to be processed. For example, in the example of
Returning to
When the character number is calculated for each speaker by the character number calculation section 62, the threshold setting section 66 may change respective thresholds of speakers independently or in conjunction with one another. For example, in the example of
A playback control section 70 controls readout of a TS from the memory means 30 and plays back audio and video data output from the AV decoder 50. In this control, the playback control section 70 performs control so that a TS corresponding to a segment specified by the digest scene specifying section 64 is read out. Thus, a digest of the television broadcast program is played back.
Playback of the digest may be performed after analyzing the entire television broadcast program which has been recorded, or may be performed while looking ahead subtitle information and specifying a segment by the digest scene specifying section 64. In the latter case, it is advantageous that only a small size buffer for temporarily storing a result of specification by the digest scene specifying section 64 is required.
As has been described, according to this embodiment, scenes of a television broadcast program each of which contains large character information by subtitles, i.e., scenes containing many conversations can be extracted throughout from the entire program. Thus, it is possible to generate an excellent digest for comprehension of a story plot of a broadcast program of drama, movie or like genre. Furthermore, by setting a threshold for each speaker, a digest in which each of persons has been weighted can be generated.
By operating the character number calculation section 62 and the digest scene specifying section 64 in parallel to recording of a television broadcast program, a digest can be generated while the television broadcast program is recorded in the memory means 30. In this case, the digest scene specifying section 64 records, in the memory means 30, a segment specified as a digest scene in association with a broadcast program being received. Thus, a digest can be immediately played back after recording of a television broadcast program. For a digest generated in the above-described manner, a playback time can not be adjusted and thus the threshold setting section 66 is not needed.
Moreover, without providing the threshold setting section 66, a threshold to be given to the digest scene specifying section 64 may be manually changed. Although a digest playback time is increased by setting the threshold to be smaller, a digest which contains a segment with a small display character number and allows easy comprehension of a story plot of the program is generated. On the other hand, by setting the threshold to be larger, a digest with a short playback time is generated. When a threshold is given for each speaker, each threshold may be manually changed.
Furthermore, symbols, external characters, parenthesized part, or the like contained in subtitles are not given for indicating a dialogue. Therefore, the character number calculation section 62 may be configured to calculate the character number so as to exclude characters other than characters describing a dialogue. Thus, a scene including utterance of each person can be extracted with high accuracy.
The digest scene specifying section 64 compares, for each segment of a television broadcast program, a character number calculated by the character number calculation section 62 with each of a plurality of thresholds. Then, the digest scene specifying section 64 specifies, for each of the thresholds, a segment in which the calculated character number is larger than an associated one of the thresholds as a digest scene of the television broadcast program to be processed.
Returning to
As has been described, according to this embodiment, when a digest playback time is designated, a digest satisfying the designated time condition can be generated in a single processing without repeating the processing of digest scene specification by changing a threshold many times. Thus, processing speed of generating a digest can be improved.
Note that, for convenience, digest generation of a digital broadcast program has been described. However, a method for generating a digest according to the present invention is applicable to teletext broadcast.
Each of the character number calculation section 62, the digest scene specifying section 64, the threshold setting section 66 and the digest selection section 68 can be realized by making computer read a program recorded in a computer-readable recording medium or a program provided through an electronic communication line and then execute the program.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2007-149703 | Jun 2007 | JP | national |
