JPH0895596A - Quick-look and quick-listening device and method thereof - Google Patents

Abstract

PURPOSE: To provide a quick-look/quick-listening device and its method capable of grasping the outer shell and the atmosphere of an image and a sound in a short time by using sound information. CONSTITUTION: Video information (information composed of picture information and sound information) or audio information is inputted in real time from a video input part 101, the video or audio information inputted in real time is stored in a video/audio feature variables storage part 105 and the stored video or audio information is inputted. Various kinds of feature variables are extracted from sound information among the inputted video or audio information by means of an audio feature extracting part 103. The video or audio information is processed based on the extracted feature variables by means of a feature variable managing part 104. Processed video or sound information is outputted from an interface part 106. Quick-look/quick-listening is realized by the processing based on the feature variables.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a quick-listening device and method suitable for grasping the contents of images and sounds in a short time.

[0002]

2. Description of the Related Art A video browser (Oba: "Motion picture list display technology: video browser",
1994 National Conference of Television Society, pp. 219-2
20, 1991). This shortens the time for viewing the video by performing an operation such as fast-forwarding the video or playing back only before and after a cut point which is a break of the video. In addition, a technology that improves the visibility of images and prints still images immediately after the cut points spatially side by side on a display or paper so that the outer shell of the images can be grasped at a glance (Tomura et al .: “Paper Video: Fixing video information on paper ", 1994 IEICE Spring National Convention, A
-315, 1994).

In the case of using sound, word spotting (Lyn Wilcox et.
dspotting for Voice Edit
ng Audio Audio Indexing, Geor
gia Institute of Technology
gy ", 1992), which involves inputting and recognizing a voice from a microphone and searching a storage medium such as a tape for a place where the same word as that word appears.

[0004]

In order to grasp the contents of images and sounds and to know what kind of images and sounds are stored in a short time, it is necessary to play or fast-forward. . However, in such a method, it takes a considerable amount of time to reach a desired portion and grasp the whole atmosphere. As a solution to these problems, the video browser as described above can be considered, but because the sound cannot be played back by simple fast-forwarding, or it depends only on the structural feature amount of the image, the selected browsing image Cannot necessarily be said to reflect what happened in the video. In addition, the sound at the joint is broken and unnatural, or there is a problem of how to select only important ones from a large number of cuts and images in order to further shorten the time. In the word spotting technique described above, the function of browsing is not considered at all because the word must be known in advance.

The present invention has been made to solve the above problems, and an object of the present invention is to provide a quick-reading speed that can grasp the outer shell or atmosphere of a picture or sound by using sound information in a short time. A listening device and method are provided.

More specifically, the image enables selection of an image or a section of sound that reflects an event that occurred in the sound, and when reproducing some sound sections, the joints between the sections are broken. We will provide a technology that makes it possible to grasp the outer shell and atmosphere of images and sounds in a shorter time by making them less natural. Also, by using sound, you can combine multiple cuts and images into several pieces, and by associating sound feature amounts with events, you can quickly see or listen according to the event you want. To provide the technology. further,
Even if the desired event is not clearly known, it is possible to periodically reproduce the sound segments corresponding to different events by exchanging them at a certain interval, which provides a technology suitable for browsing.

[0007]

In order to achieve the above object, in the invention of claim 1, a video / sound information input section for inputting video information consisting of image information and sound information or sound information in real time is provided. An image / sound / feature amount storage unit that stores the input image or sound information in real time, outputs the stored image or sound information, and stores a feature amount of the image or sound information.
An image / sound management control unit that extracts various feature amounts from sound information in the input image or sound information and manages the feature amounts, and an interface unit that outputs the image or sound information in which the feature amounts are managed. And a quick-listening and listening device having

According to the second aspect of the present invention, the video / sound management control section extracts the various feature quantities from the sound information, the sound feature extraction section manages the feature quantities, and outputs the sound information based on the feature quantities. A fast-viewing / listening device having a feature amount management unit for processing and converting is used as means.

According to a third aspect of the present invention, the sound feature extraction unit includes sound information change detecting means for extracting various feature quantities from sound information and detecting from the sound information a sound section in which a feature quantity satisfying a certain condition exists. The fast-listening and listening device provided is used as means.

According to a fourth aspect of the present invention, there is provided a quick-speed listening device in which the feature quantity management section includes event associating means for associating a sound section having a certain feature quantity with an event occurring in a video or sound. Use it as a means.

According to a fifth aspect of the present invention, the feature amount management unit is a fast-speed listening device that includes a joining unit that joins the sound sections having a particular feature amount or corresponding to a particular event. .

In a sixth aspect of the present invention, the feature quantity managing section is a fast-speed listening device having a reproducing means for synchronizing image information corresponding to the sound section when the input is video information.

According to the invention of claim 7, the sound information change detecting means calculates a frequency spectrum from the sound information, detects a peak from the envelope of the spectrum, and selects a sound section satisfying a condition that the feature quantity of the peak satisfies a certain condition. A quick-listening and listening device that is detected from sound information is used.

In the invention of claim 8, the event associating means associates the sound segment with the event by comparing the feature quantity of the peak with the feature quantity of the peak in a typical event. A fast-listening device is used.

According to a ninth aspect of the present invention, the connecting means connects the sound sections having a specific feature quantity or event included in the sound information and reproduces the total sum of the time required for reproducing the feature quantity or the event. A quick-reading speed listening device having sound section selection means for selecting the sound section closest to the time limit according to the limitation of the reproduction time is used as the means.

According to a tenth aspect of the present invention, the connecting means is a quick-speed listening device in which the connecting point is a point where the power of the sound information is low in the selected sound section. .

According to an eleventh aspect of the invention, there is provided a fast-speed listening device in which the joining means lowers the power at both ends of the selected sound section to be a joining point. To do.

In a twelfth aspect of the present invention, the reproducing means is
Fast-speed listening with a means for reproducing a set of sound sections classified by a sound information change extracting means by a certain characteristic amount or a specific event associated by a connecting means by periodically changing each amplitude. The device is the means.

In the thirteenth aspect of the present invention, first, image information or sound information consisting of image information and sound information is input in real time, and the image or sound information input in the real time is accumulated. In the input video or sound information, various feature quantities are extracted from the sound information in the input video or sound information, and then the video or sound information is extracted based on the extracted feature quantities. Process sound information,
Next, a quick-listening and fast-listening method for outputting the processed image or sound information is used as means.

According to the fourteenth aspect of the invention, in the process of processing the image or the sound information, a sound section in which a feature quantity satisfying a certain condition is detected from the sound information among various feature quantities extracted from the sound information. The method is a fast-viewing and fast-listening method using the sound information change detection method.

The fifteenth aspect of the present invention uses a fast-speed listening method that uses an event association method for associating a sound section having a certain characteristic amount with an event that occurs in a video or sound.

In the sixteenth aspect of the present invention, a fast-listening fast listening method using a joining method for joining the sound sections having a particular feature quantity or corresponding to a particular event is used.

In the seventeenth aspect of the present invention, when the input is video information, a fast-viewing speed listening method using a reproducing method for synchronizing image information corresponding to the sound section is used.

According to an eighteenth aspect of the present invention, in the sound information change detecting method, a frequency spectrum is calculated from sound information, a peak is detected from an envelope curve of the spectrum, and a feature quantity of the peak satisfies a certain condition. A quick-listening and fast-listening method using a method of detecting a section from sound information is used.

According to a nineteenth aspect of the present invention, there is provided the method of associating an event, wherein the feature amount of the peak is compared with the feature amount of the peak in a typical event to associate the sound segment with the event. Use the fast-seeing, fast-listening method used.

According to a twentieth aspect of the present invention, there is provided the above-mentioned joining method, wherein a total time required for joining and reproducing sound sections having a specific feature amount or event included in sound information is used as the feature amount or A quick-reading speed listening method that uses a sound segment selection method that calculates for each event and selects a sound segment that is closest to the time limit in accordance with the reproduction time limit is used.

According to a twenty-first aspect of the invention, there is provided a fast-connecting fast listening method using the joining method, wherein a joining point is a portion where the power of sound information is low in the selected sound section. And

According to a twenty-second aspect of the present invention, there is provided a fast-connecting fast listening method, which is the joining method, wherein a method is used in which the power at both ends of the selected sound section is lowered to form a joining point. Use it as a means.

According to the twenty-third aspect of the present invention, a set of sound sections classified for each specific event associated with a certain feature amount or a joining method by the sound information change extraction method changes each amplitude periodically. A fast-viewing and fast-listening method using a reproduction method of reproducing by making it is a means.

[0030]

According to the fast-viewing speed listening apparatus and method of claims 1 and 2 and 13 of the present invention, video information (information consisting of image information and sound information) and sound information are input in real time, and input in real time. The image or sound information is accumulated, the accumulated image or sound information is input, various characteristic quantities are extracted from the sound information in the inputted image or sound information, and the image or sound information is extracted based on the characteristic quantity. By using the means for managing / processing and outputting the managed / processed video or sound information, the video or sound information is input regardless of real-time information or accumulated information,
It manages / processes, outputs, and manages / processes based on the feature quantity to realize quick-seeing and quick-listening.

According to the third aspect of the present invention, the sound feature extracting section extracts a variety of feature quantities from the sound information, and detects a sound section in which the feature quantity satisfying a certain condition exists from the sound information. Enable to divide sound information.

According to the fourth aspect of the present invention, the feature quantity management unit can classify the sound information by using means for associating a sound section having a certain feature quantity with an event occurring in the image or the sound. To do.

The feature quantity management unit according to claim 5 of the present invention collects desired sound sections by using means for connecting the sound sections having a particular feature quantity or corresponding to a particular event.

According to the sixth aspect of the present invention, when the input is video information, the feature quantity management unit uses means for synchronizing the image information corresponding to the sound section, thereby reproducing the image and the sound without deviation. .

According to claim 14 of the present invention,
The sound information can be divided by extracting various feature quantities from the sound information and detecting the sound section in which the feature quantity satisfying a certain condition exists from the sound information.

According to a fifteenth aspect of the present invention, there is provided a fast-viewing and fast-listening method,
The sound information can be classified by using a method of associating a sound section having a certain characteristic amount with an event occurring in a video or a sound.

According to claim 16 of the present invention,
A desired sound section is put together by using a method of connecting the sound sections having a specific characteristic amount or corresponding to a specific event.

According to a seventeenth aspect of the present invention, there is provided a quick-listening and fast-listening method,
When the input is video information, a method of synchronizing the image information corresponding to the sound section is used to reproduce the image and the sound without displacement.

According to the sound information change detecting means and method of the present invention, the frequency spectrum is calculated from the sound information, the peak is detected from the envelope of the spectrum, and the feature quantity of the peak satisfies a certain condition. By using the method of detecting the section of the sound from the sound information, the change of the sound is detected and the feature is detected.

The event associating means and method according to claims 8 and 19 of the present invention associates the sound segment with the event by comparing the feature quantity of the peak with the feature quantity of the peak in a typical event.

The connecting means and method according to claims 9 and 20 of the present invention is to connect the sound sections having a specific characteristic amount or event included in the sound information, and to sum the total time required for reproduction, Alternatively, the reproduction time can be changed by using a means for calculating for each event and selecting the sound section closest to the time limit in accordance with the restriction of the reproduction time, and the quick-listening listening according to the time limit is realized.

In the connecting means and method according to claims 10 and 21 of the present invention, when the selected sound sections are connected by using the method of using the connecting point where the power of the sound information is low. Eliminates the unnatural breaks in the sound that occur in.

The connecting means and method according to the eleventh and twenty-second aspects of the present invention uses the method of lowering the power at both ends of the sound section to use as a connection point, thereby connecting the selected sound sections. Eliminates unnatural breaks in the sound that occur at the time.

According to the twelfth and the twenty-third reproduction means and methods of the present invention, the sound information change extraction means classifies the sound features classified into a certain feature amount or a specific event associated by the connection means. By using a means to periodically change the amplitude of each group, sounds corresponding to different features and events are played in sequence at a certain interval, and it is possible to quickly determine what features or events are included. Can be grasped with.

[0045]

An embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of a fast-viewing speed listening apparatus according to an embodiment of the present invention. The fast-viewing speed listening apparatus of this embodiment includes a video / sound input unit 101 for inputting video or sound information, and a video / sound / feature amount storage unit 105 for sequentially storing video, sound, and feature amounts input in real time. And the video,
Video / sound management control unit 10 that manages and controls sounds and feature quantities
2 and an interface unit 106 that presents the output video and sound. Video / sound management control unit 10
2 is a sound feature extraction unit 103 that extracts a feature of sound information;
The sound feature extraction unit 1 includes a feature amount management unit 104 that processes, converts, and reproduces video or sound based on the feature amount.
03 and the feature amount management unit 104 operate in parallel or in time division, and can process, convert, and reproduce video or sound while extracting features in real time.

FIG. 2 is a block diagram showing the arrangement of the video / sound management control unit 102. The sound feature extraction unit 103 includes a sound information change detection unit 201 that detects a sound section that satisfies a certain condition of the sound, and the feature amount management unit 104 includes a sound section obtained by the sound information change detection unit 201. It is composed of an event associating means 202 for associating an event occurring in a video or a sound, a connecting means 203 for connecting the sound sections, and a reproducing means 204 for synchronizing image information corresponding to the sound sections.

FIG. 3 is a flowchart showing the processing when the sound feature extraction unit 103 and the feature amount management unit 104 of the video / sound management control unit 102 are realized by software such as a computer. In this case, first, the sound information change detection processing 301
Then, event association processing 302 is performed, and then
The joining process 303 is performed, and then the reproduction process 304
I do.

Next, the operation of this embodiment will be described.

FIG. 4 is a flow chart showing the processing of the sound information change detecting means 201 or the sound information change detecting processing 301,
5 is a diagram showing a state in which a frequency spectrum and a spectrum envelope are calculated from a sound waveform according to the present embodiment, FIG. 6 is a diagram showing peaks and feature amounts of the spectrum envelope according to the present embodiment, and FIG. 7 is a present embodiment. FIG. 8 is a diagram showing the time change of the power of the waveform used in the connecting means according to the present invention, and FIG. 8 is a diagram showing the weighting function used in the reproducing means according to this embodiment.

The video / sound information is sent to the video / sound input unit 101.
When the video or sound is input in real time, the video / sound / feature amount storage unit 105 sequentially stores the video or sound. The input video information is separated into image information and sound information, and the sound information is analyzed by the sound information change detecting means 201. When the input is only the sound information, it is directly analyzed by the sound information change detecting means 201. In the sound information change detecting means 201, the spectrum calculation processing 401 of FIG. 4 is first performed. The waveform of the input sound information is 5 in FIG.
If it is given as 01, it will be a few ms like 502.
It is cut out into a frame of several tens of ms and subjected to FFT (Fast Fourier Transform) processing. The waveform subjected to the FFT processing has a large undulation in the power direction as indicated by 503, and is not suitable for peak detection. Therefore, the frequency spectrum envelope calculation process 402 of FIG. 4 is performed to obtain the FFT cepstrum, the lifter process is performed, and then the FF is performed again.
The T process is performed to calculate the envelope of the spectrum. The calculated envelope is 504. As a method for calculating the cepstrum, a method using LPC (linear prediction analysis) in addition to FFT can be considered.

Next, from the obtained envelope curve, 403 in FIG.
The peak detection process shown in is performed. Since the envelope is smoothed by the lifter process, the peak can be easily obtained from the maximum value and the minimum value of the envelope. Examples of the obtained peaks are shown in 604 to 608 of FIG.

Frames cut out from the waveform of sound information are several meters
The same process is repeated by shifting in the time axis direction for about s to several tens of ms. In this way, the peak variation calculation processing 404 is performed based on the peaks obtained at each time. In this process, changes with time of various parameters based on the peak are obtained. As the parameters, the peak power 601, the peak frequency 602, and the power difference 603 from the maximum peak to the adjacent minimum peak shown in FIG.
The number of peaks is considered.

Based on the temporal changes in these parameters, the event associating means 202 associates with the event that occurred in the image or sound. For example, if the image or sound contains music, the peak frequency 602 varies with time with a threshold value or less, or the peak power 6 is exceeded at a constant cycle.
Since a feature such as 01 becoming larger can be seen, a sound section including such a feature is associated with music. In addition, when the frequency of the peak changes smoothly, or when features such as a large difference between the maximum peak and the minimum peak are seen, it is associated with the human voice, and when the power or number of the peaks fluctuates finely, It is associated with crowds and noise.

Next, by the connecting means 203, for example, when it is desired to reproduce only the human voice section, the sound section of the human voice correlated by the event correlating means 202 is selected and connected. At this time, if 702 in FIG. 7 is one of the selected sound sections, 703, 704
By connecting the portions where the power is low in the vicinity of the beginning and end of the sound section as described above, it is possible to reduce the unnatural break of the sound. Also, the sound section 702
There is also a method of connecting by forcibly lowering the power at both ends of. For the calculation of power, a method of dividing the sound waveform within a range of several ms and using the sum of squares of the amplitude can be considered. An example of the calculated power is shown at 701.

The sound section to be reproduced is not limited to the connection of sections having a fixed characteristic amount, but may be a fixed time before and after the point at which the sound information changes abruptly.

The collection of sound sections generated by the connecting means 203 is reproduced by the reproducing means 204. At this time, when the input information is a video, it is reproduced together with an image corresponding to each sound section. When the input information is only sound, it is reproduced as it is.

The reproducing means also has the following browsing function. 801 to 8 of FIG. 8 are arranged in the amplitude direction in the sound sections connected by a certain feature amount or each specific event.
Periodically changing weighting such as 04 is applied, and a phase difference as indicated by 805 is given for reproduction. FIG. 8 shows the case where there are four sound segments, but the number of sound segments, the weighting function, the phase difference, etc. can be set arbitrarily. By reproducing by such a method, the sound sections corresponding to different features or events are heard while being switched.

[0059]

As described above, the present invention has the following effects.

According to the first to twelfth aspects of the present invention, the feature is detected from the sound information, and the feature is correlated with the event that occurred in the image or the sound, and the desired sound segment or the sound segment satisfying a certain condition is satisfied. Since it is reproduced, it is possible to perform a quick viewing or a quick listening reflecting the contents of the image or the sound.

The connecting means method according to claims 9 and 20 connects the sound sections having a specific feature amount or event included in the sound information, and sums the time required for reproduction for each feature amount or each event. Since the calculated sound section is selected and the sound section closest to the reproduction time limit specified by the user is selected, it is possible to perform quick-reading or quick listening according to the reproduction time limit.

In the connecting means and method according to the tenth and the twenty-first aspects of the present invention, since the part where the power of the sound information becomes low when connecting the selected sound sections is the connecting point, it is unnatural. It is possible to reduce the break in the sound.

According to the connecting means and method of claims 11 and 22 of the present invention, when connecting the selected sound sections, the power of both ends of the sound sections is lowered so that the selected sound sections are connected. Cuts can be reduced.

The reproducing means and method according to the twelfth and twenty-third aspects of the present invention reproduces a set of sound sections classified by a certain characteristic amount or a specific event by periodically changing the amplitude of each. It becomes possible to grasp in a short time what kind of characteristic amount or event is included.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a fast-viewing speed listening apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a video / sound management control unit according to an embodiment of the present invention.

FIG. 3 is a flowchart showing a processing flow when the operations of the sound feature extraction unit and the feature amount management unit according to the embodiment of the present invention are realized by software such as a computer.

FIG. 4 is a flowchart for explaining the flow of processing from the frequency spectrum calculation processing of sound information to the calculation of the peak variation of the envelope of the frequency spectrum according to the present embodiment.

FIG. 5 is a diagram showing an example of a waveform of sound information according to the present embodiment, a frequency spectrum calculated by FFT, and a state of an envelope of a frequency spectrum calculated by FFT after performing a lifting process from the cepstrum.

FIG. 6 is a diagram showing an envelope of a frequency spectrum of sound information and various parameters used to detect changes in the sound information according to the present embodiment.

FIG. 7 is a diagram showing a state of power of sound information according to the present embodiment and a joint portion where the power is low.

FIG. 8 is a diagram showing a state of a weighting function of a sound section and a phase difference according to the present embodiment.

[Explanation of symbols]

 101 ... Video / sound input section 102 ... Video / sound management control section 103 ... Sound feature extraction section 104 ... Feature amount management section 105 ... Video / sound / feature amount storage section 106 ... Interface section 201 ... Sound information change detection means 202 ... event associating means 203 ... connecting means 204 ... reproducing means 301 ... sound information change detecting processing 302 ... event associating processing 303 ... connecting processing 304 ... reproducing processing 401 ... frequency spectrum calculating processing 402 ... frequency spectrum envelope calculating processing 403 ... Peak detection processing 404 ... Peak fluctuation calculation processing

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H04N 5/928

Claims (23)

[Claims] 1. A video / sound information input unit for inputting video information or sound information consisting of image information and sound information in real time, and storing the video or sound information input in the real time. An image / sound / feature amount storage unit that outputs image or sound information and accumulates the feature amount of the image or sound information, and various feature amounts by extracting various feature amounts from the sound information in the input image or sound information. And a sound / speed management device that manages the video and sound, and an interface unit that outputs video or sound information in which the characteristic amount is managed. 2. The video / sound management control unit, a sound feature extraction unit that extracts various feature amounts from sound information, and a feature that manages the feature amount and processes and converts the sound information based on the feature amount. The quick-listening listening device according to claim 1, further comprising a quantity management unit. 3. The sound feature extraction unit includes sound information change detecting means for extracting various feature quantities from sound information and detecting from the sound information a sound section in which a feature quantity satisfying a certain condition exists. The rapid-watching and listening device according to claim 2. 4. The feature quantity management unit comprises event association means for associating a sound section having a certain feature quantity with an event occurring in a video or a sound. Quick-speed listening device. 5. The feature quantity management unit comprises a joining means for joining the sound sections having a particular feature quantity or corresponding to a particular event, to each other. Quick-listening device. 6. The feature quantity managing section comprises a reproducing means for synchronizing the image information corresponding to the sound section when the input is video information. Item 10. A quick-listening and fast-listening device according to Item 5. 7. The sound information change detecting means calculates a frequency spectrum from the sound information, detects a peak from the envelope of the spectrum, and detects a sound section in which the feature amount of the peak satisfies a certain condition from the sound information. The quick-listening fast-listening device according to claim 3, wherein 8. The event associating means associates the sound segment with an event by comparing the feature value of the peak with the feature value of the peak in a typical event. Item 10. A quick-listening and fast-listening device according to Item 4. 9. The connecting means connects sound sections having a specific feature amount or event included in the sound information, and calculates the total time required for reproduction for each feature amount or each event, 6. The fast-paced speed listening device according to claim 5, further comprising a sound section selecting unit that selects a sound section closest to the time limit in accordance with the reproduction time limit. 10. The connecting means according to claim 5, wherein the connecting point is a connecting point when the power of the sound information is low in the selected sound section. Quick-listening device. 11. The connecting means for lowering the power at both ends of the selected sound section as a connecting point in the selected sound section. Fast-listening and listening device described. 12. A reproduction for reproducing a set of sound sections classified by a certain feature amount by the sound information change extracting means or a specific event associated by the connecting means by periodically changing each amplitude. The quick-speed listening device according to claim 4, 5, or 6, further comprising means. 13. First, image information or sound information consisting of image information and sound information is input in real time, and the image or sound information input in said real time is accumulated, and if it is not in real time, it is accumulated. Inputting image or sound information, then extracting various feature amounts from the sound information in the input image or sound information, and then processing the image or sound information based on the extracted feature amount. Next, a quick-listening and fast-listening method, characterized in that the processed image or sound information is output. 14. A sound information change detection for detecting, from the sound information, a sound section in which a feature quantity satisfying a certain condition among various feature quantities extracted from the sound information is detected in the process of processing the image or the sound information. 14. The method for listening to fast-paced speed according to claim 13, wherein the method is used. 15. The rapid-speed listening method according to claim 14, wherein an event association method is used for associating a sound section having a certain characteristic amount with an event occurring in a video or a sound. 16. The method according to claim 14 or 1, wherein a joining method for joining the sound sections having a particular feature quantity or corresponding to a particular event is used.
The quick-listening and listening method described in 5. 17. The reproducing method for synchronizing the image information corresponding to the sound section when the input is video information, according to claim 14 or claim 15 or claim 1.
The quick-listening and listening method described in 6. 18. The sound information change detecting method, wherein a frequency spectrum is calculated from sound information, a peak is detected from an envelope of the spectrum, and a sound section in which a feature amount of the peak satisfies a certain condition is detected from the sound information. 15. The fast-listening fast-listening method according to claim 14, wherein a detecting method is used. 19. The method of associating an event, wherein a method of associating the sound segment with an event by comparing the feature value of the peak with the feature value of the peak in a typical event is used. The quick-listening, fast-listening method according to claim 15. 20. The connection method, wherein a total sum of time required for reproducing and reproducing a sound section having a specific feature amount or event included in sound information is calculated for each feature amount or each event. 17. The fast-listening fast listening method according to claim 16, wherein a sound segment selection method for selecting a sound segment closest to the time limit is used according to the limit of the reproduction time. 21. The joining method, wherein the joining point is a portion where the power of sound information is low in the selected sound segment. 20. Quick listening and listening method described in 20. 22. The joining method, wherein in the selected sound section, a method is used in which the power at both ends of the sound section is lowered to form a joining point. Item 20. A quick-listening and listening method according to Item 20. 23. A reproduction for reproducing a set of sound sections classified by a sound information change extraction method by a certain feature amount or a specific event associated by a joining method by periodically changing each amplitude. The method of using the method of claim 15 or claim 16 or claim 17, characterized in that the method is used.