JPH10312389A - Voice data base system and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain voice data that an operator desires speedily, efficiently, and securely. SOLUTION: Key phrases are set for all voice data registered in the voice data base system. A key phrase is sectional voice information on a featured part of a voice that voice data indicate. Then the voice data base system retrieves voice data under arbitrary conditions. For all voice data extracted by the retrieval, voice waveforms WAV4 and WAV5 are generated by repeating key phrases. Then a voice waveform WAV6 is synthesized by putting together all voice waveforms having key phrases repeated. Further, the level of the voice waveform WAV6 is adjusted to generate a vice waveform WAV7. When the retrieval result is displayed, a voice corresponding to the voice waveform WAV7 is reproduced to make it easy to specify the voice data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio database for storing a large amount of audio data in a computer that handles audio data, searching for desired audio data from the stored audio data, and displaying and reproducing the extracted audio data. The present invention relates to a system and a recording medium.

[0002]

2. Description of the Related Art When a large amount of voice data is stored and managed in a computer and used, a voice database system capable of more efficient retrieval and quick confirmation is required as the number of stored data increases.

[0003] However, since voice data is data relating to voice recognized in hearing, the operator cannot visually recognize the data simply by displaying the content of the data on a display device like text data or image data. .

Therefore, in the conventional voice database system, attribute information such as a keyword is associated with voice data in order to perform an efficient search, and a search is performed from a large amount of voice data based on the attribute information. It is configured as follows. The attribute information includes a title name, a creator / recorder, a recording location, a recording date, a recording time, and the like in addition to the keyword. Then, the audio data extracted as a result of the search is displayed on the screen of the display device in a list format.

FIG. 16 is an explanatory diagram showing an operation procedure in such a conventional voice database system on a screen displayed on a display device. As shown in FIG. 16, the operator inputs attribute information such as a keyword into the search condition input area 25 of the search screen P1 in order to obtain desired voice data. For example, when the operator requests voice data of a cat's meow, a word such as "cat" is input as a keyword, and the computer is caused to execute a search.

[0006] When the computer completes the search, the search result is displayed on the display device. The search result display screen P2 shown in FIG. 16 shows a case where there are six extracted voice data. In the search result display screen P2, audio data is displayed by "icon" and "file name". Here, the “file name” is a name associated with data or a program, and the “icon” is a pictorial symbol for displaying a file as a symbol on a screen. Then, a picture of a speaker is adopted as an icon indicating audio data.

[0007]

However, in general, the operator cannot often determine the file name set by another person, and the search result is displayed as described above. Even if the operator does not recognize the file name, the final audio data cannot be specified.

[0008] Therefore, in order for the operator to determine the desired voice data, it is necessary to display the attribute information and confirm it in detail. Identifying audio data while checking attribute information in this way is not an easy task, and is extremely inefficient,
Since no actual sound reproduction is involved, the operator may sometimes feel uneasy.

[0009] If the operator remains uneasy by checking the attribute information, the operator activates a voice data reproduction application and reproduces the actual voice. In this operation, the reproduction application must be intentionally started from the search result display screen P2, and further, the operation on the reproduction application screen P3 must be performed, and the operation is troublesome. Further, when the audio data is long, even if the audio data is reproduced, it may take a long time to reach a characteristic portion of the audio, and efficient identification cannot be performed.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has as its object to provide a voice database system and a recording medium capable of quickly and efficiently obtaining voice data desired by an operator. I do.

[0011]

In order to achieve the above-mentioned object, the invention according to claim 1 is a computer which handles audio data, comprising: Storage means for storing section voice information in association with the voice data, (b) a search means for searching for voice data based on predetermined information from a plurality of voice data, and (c) extraction by the search means Output means for reproducing the voice data obtained based on the section voice information and outputting the voice data as voice.

According to a second aspect of the present invention, in the system according to the first aspect, the storage means can store a plurality of section audio information in association with one audio data. I have.

According to a third aspect of the present invention, in the system according to the first or second aspect, the display corresponding to the audio data output by the output means is accompanied by reproduction based on the section audio information of the audio data. It is characterized by changing.

According to a fourth aspect of the present invention, in the system according to any one of the first to third aspects, for the voice data extracted by the search means, the onomatopoeia of the pronunciation of a characteristic portion or the voice data is used. And a voice data recognition word expressed in characters including the objective object of the present invention.

According to a fifth aspect of the present invention, there is provided a storage means for storing (a) section voice information on an arbitrary characteristic portion of voice indicated by voice data in association with the voice data; A) searching means for searching for audio data based on predetermined information from a plurality of audio data, (c)
A voice database program is recorded for reproducing the voice data extracted by the search means based on the section voice information and for functioning as an output means for outputting as voice.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION

<1. Configuration of Apparatus> First, an outline of a voice database system according to an embodiment of the present invention will be described.
FIG. 1 is a schematic diagram showing a configuration of a voice database system according to an embodiment of the present invention. As shown in FIG. 1, in this device, an input / output device 11, a CPU 12, a memory 13, a storage unit 14, interfaces 15, 16, 1
7 are mutually connected via a bus line BL. The input / output device 11 is a device that reads data from and writes data to a computer-readable portable recording medium D such as a flexible disk, a magneto-optical disk, and a CD-ROM. CPU 12
Is a processing unit that performs arithmetic processing. The memory 13 is a device for temporarily storing and holding data, and the storage unit 14 is a computer-readable fixed recording medium such as a magnetic disk. And interface 1
A display device 18 such as a CRT or a liquid crystal display is connected to 5, and a keyboard 19 and a mouse 20 are connected to the interface 16. Further, a speaker 21 for generating audio for audio data is connected to the interface 17. The voice database system can be connected to a network as needed, and can obtain voice data from other devices connected to the network.

As described above, the voice database system according to this embodiment is a device realized by executing the voice database program by the internal CPU 12 in one general computer. Note that the above audio database program may be read from the portable recording medium D, or may be stored in the storage unit 14 in advance. That is, the object in which the audio database program is stored may be a portable recording medium or a fixed recording medium.

<2. Registration of Voice Data> In the voice database system of this embodiment, when registering voice data, the operator sets attribute information and a voice data recognition word. As information included in the attribute information,
The title, creator / recorder, recording location, recording date, recording time, keyword, comment, etc., of the audio data are provided. Note that a plurality of keywords can be set. This attribute information is information used when searching for audio data. Also, "Speech data recognition word"
Is an arbitrary word that specifically and simply indicates what kind of sound the sound data is. For example, if the sound data is the sound of a bell
A speech data recognition word such as an onomatopoeic word “Lean Lean”, a specific name “Suzumushi cry”, and a name of an objective object “Suzumushi” as a source of the sound is input. Since the voice data recognition word is character data used for displaying search results, it is preferable to input the content of the voice data using words such as a song title or an onomatopoeia that can be recognized by anyone.

The operator inputs the attribute information and the voice data recognition word from the keyboard 19 or the mouse 20. The input attribute information and voice data recognition word are stored in the storage unit 14. When a large amount of audio data is stored in a recording medium such as a CD-ROM, etc., the attribute information, the audio data recognition word, and the The data recognition word is stored in the storage unit 14. As a result, there is no need to store audio data requiring a large capacity in the storage unit 14.

When registering voice data, the operator sets a key phrase of the voice data. The “key phrase” is section voice information of an arbitrary characteristic portion of the voice indicated by the voice data. Note that it is also possible to set a plurality of key phrases for one audio data, and by setting a plurality of key phrases, the identification of the audio data becomes more reliable. The setting of the key phrase is performed by inputting a start point and an end point of a distinctive characteristic portion of the audio data.

FIG. 2 is an explanatory diagram for explaining a method of setting a key phrase of audio data in this embodiment. As shown in FIG. 2, when the reproduction of the audio data starts at time t1, the process ends at time t2. In FIG. 2, a sound waveform WAV1 shows a sound waveform when sound data is reproduced. Such an audio waveform WAV1
Is displayed on the display device 18 at the time of setting the key phrase so that the audio data can be visually recognized, and the reproduction position display 22 changes from the position of the time t1 to the position of the time t2 with the reproduction of the audio data. It is configured to move up to. Then, assuming that the characteristic part of the audio data is the audio between time ta and time tb, the operator inputs the start point from the keyboard 19 or the like when approaching the time ta after starting to reproduce the audio data. At the time tb, the keyboard 19
Input the end point from In this way, for the voice data, a characteristic portion of the voice between time ta and tb is set as a key phrase.

In the setting of such a key phrase, a plurality of designations can be made for one voice data. When a plurality of key phrases are set, the input of the key phrase section between the start point and the end point may be repeated.
If the key phrase is not set, all sections of the audio data are automatically set as the key phrase. The key phrase is also stored in the storage unit 14 in a state where the key phrase is associated with the attribute information, the voice data recognition word, and the voice data. This storage unit 1
The storage in 4 may store information indicating the start point and end point of the key phrase, or may extract and store the actual voice data of the section set as the key phrase.

Next, a procedure for registering voice data in the voice database system will be described. FIG. 3 is a flowchart showing a process of registering audio data in the audio database system according to this embodiment.

First, in the CPU 12, a program relating to registration of voice data among voice database programs is started in accordance with an instruction of an operator (step S).
11). Next, in step S12, the operator
Unregistered audio data is selected from among audio data held in a storage medium such as a D-ROM or the storage unit 14. This selection is made by the operator while referring to a list of audio data displayed on the display device 18. And step S
At 13, the CPU 12 reproduces the audio data selected by the operator and causes the speaker 21 to generate audio.
Then, the operator listens to the voice from the speaker 21 and determines whether or not the voice data is to be registered in the voice database system.
S ”or“ NO ”(step S
14). If the registration process is to be advanced by the input of the operator in step S14, step S15
Proceeds to step S12 if the registration process is not performed.
And the other audio data is selected.

In step S15, input is made of attribute information of voice data to be registered and voice data recognition words. This input is also performed by the operator using the keyboard 19 or the like while referring to the display device 18. And then
The key phrase is set (step S16). As described above, the setting of the key phrase is performed by inputting the start point and the end point of the characteristic portion of the voice data based on the content displayed on the display device 18 and the voice heard from the speaker 21. Then, in step S17, regarding the contents input and set in steps S15 and S16,
The operator issues a registration instruction. Upon receiving the registration instruction, the CPU 12 associates the voice data, the attribute information, the voice data recognition word, and the key phrase with each other, and
4. Save the contents. Then, in step S18, the operator determines whether or not to end the registration processing, and performs an input corresponding to each. The CPU 12 advances the process to step S12 or step S19 based on the input in step S18. Then, in step S19, the CPU 12 terminates the program relating to the registration of the audio data.

With the above-described processing, the voice database system of this embodiment can set attribute information, voice data recognition words, and key phrases for voice data.

<3. Search for Voice Data and Display of Search Results> Next, search for voice data and display of search results will be described.

In the search for audio data, a search based on attribute information such as a title name, a keyword, and a comment or a file name is performed in the same manner as a conventional search. For example, if the operator requests voice data of a bark of a cat, inputting a search word such as "animal" or "cat" at the time of search will obtain voice data having such a search word in attribute information or a file name. be able to. Also, search is CPU
12 is performed based on the attribute information and the like stored in the storage unit 14, and the audio data associated with the registration can be specified.

The audio data obtained as a result of the search is displayed on the display device 18. FIG. 4 is a diagram showing an example of display of a search result in the audio database system according to the present embodiment. As shown in FIG. 4, the icons IC1 to IC4 of the four voice data are displayed on the search result display screen P2.
C4 is displayed. And each icon I
Voice data recognition words are displayed in the lower columns of C1 to IC4. "Lean Lean" is displayed on the icon IC1.
The icon IC2 displays “Wanwan”, the icon IC3 displays “Stone Rolls”, and the icon IC4 displays “Polonise” as voice data recognition words. When the voice data is registered, the voice data recognition word is registered in words that make it easy to determine what kind of voice the voice data is. Therefore, the search result display screen P2 shown in FIG. It is easy to estimate what kind of voice each of the four voice data is.

Further, in this embodiment, in order to facilitate the specification of the voice data desired by the operator, the display state of the search result based on the two states of "non-selection state" and "temporary selection state" And the sound generated from the speaker 21 is controlled. The setting of the non-selection state and the provisional selection state can be performed for each of the four voice data displayed as shown in FIG. 4, for example. The switching between the non-selection state and the provisional selection state is performed by moving the mouse pointer to a position overlapping the icons IC1 to IC4 by operating the mouse 20 or the like, and clicking the mouse 20 at that position. it can. When the mouse pointer is positioned at a position overlapping the icon of the audio data in the non-selected state, the audio data corresponding to the icon is in the tentatively selected state. Thus, intentional switching by the operator and dynamic switching coexist. FIG. 5 is a diagram showing a non-selected state and a provisionally selected state of the search result display screen P2. Icons IC1, IC3, and IC4 shown in FIG. 5 indicate audio data in a non-selected state, and icon IC2 indicates audio data in a tentatively selected state. That is, when the operation is temporarily selected by the operation of the mouse 20 or the like, the icon frame is displayed as a thick frame.

The "non-selected state" and the "temporarily selected state" will be described.

A) Control of Display in Non-Selected State When displaying search results, a case where all audio data that matches the search condition is in the non-selected state will be described. In this case, two reproduction modes “mode 1” and “mode 2” are prepared, and any mode can be selected in the initial setting of the audio database system.

A-1) Mode 1 In the "mode 1", the section of the key phrase set for each of the voice data that matches the search condition is repeated, and the repeated key phrase of all the voice data is deleted. This is a function of generating a voice synthesized by the speaker 21 after performing total synthesis, adjusting the level so that the volume becomes half of the average, and reproducing. This is shown in FIG.
This will be described with reference to FIG.

FIG. 6 is a diagram showing key phrases of two voice data. The key phrase set for the audio waveform WAV2 of the audio data shown in FIG. 6A is audio data in the range of the section ka. The key phrase set for the audio waveform WAV3 of the audio data shown in FIG. 6B is audio data in the range of section kb.

FIG. 7 is an explanatory diagram showing the summation of key phrases. The audio waveform WAV4 shown in FIG.
The key phrase section ka of the audio waveform WAV2 shown in FIG.
Is shown repeatedly. Also, the audio waveform W shown in FIG.
AV5 indicates the repetition of the key phrase section kb of the audio waveform WAV3 shown in FIG. If a plurality of key phrases are set in one audio data, they are sequentially repeated. If there are other voice data extracted as a result of the search, voice data in which the key phrase section is repeated is generated for those voice data. When each key phrase is repeated, the level of the connection between the key phrases is adjusted,
Fade-in and fade-out effects are applied.
"Fade in" means gradually increasing the audio level at the beginning of a key phrase,
"Fade-out" refers to gradually lowering the audio level at the end of a key phrase.

Then, the voice data obtained by repeating the key phrases of all the voice data of the voice data extracted by the search are summed and synthesized to generate an averaged voice waveform WAV6. Then, the audio level of the averaged audio waveform WAV6 is adjusted so as to be “１ ／”, and an audio waveform WAV7 is generated. The audio waveform WAV7 generated by performing the total synthesis, averaging, and level adjustment of all the audio data extracted by the CPU 12 in this manner.
Is reproduced, and the speaker 21 generates sound based on the sound waveform WAV7.

As described above, in the "mode 1", when the number of audio data extracted as a result of the search is small, it is determined which audio data is searched by the audio waveform WAV.
7 can be estimated by voice. Also,
When the number of voice data extracted as a result of the search is large, the key phrases of all voice data are reproduced simultaneously, and the voice is close to noise. The display screen in “mode 1” is the same as that shown in FIG.

A-2) Mode 2 In the "mode 2", the key phrase sections set for all the audio data extracted in accordance with the search condition are sequentially continued, and a half This is a function of reproducing a sound after adjusting the level so as to have a volume of, and generating a synthesized voice from the speaker 21. This will be described with reference to FIG.

FIG. 8 is an explanatory diagram of the reproduction mode of the audio data in "mode 2", and illustrates a case where the number of audio data extracted as a result of the search is three. FIG.
The key phrase set for the audio data shown in (a) is audio data in the range of the section ka, and FIG.
The key phrase set for the audio data shown in (b) is audio data in the range of section kb, and is shown in FIG.
The key phrase set for the audio data shown in (c) is audio data in the range of section kc.

In order to reproduce the key phrases shown in FIGS. 8A to 8C sequentially and continuously, FIG.
An audio waveform WAV8 shown in (d) is generated. Sound waveform W
The AV8 is simply a series of the respective key phrase sections ka, kb, kc shown in FIGS. 8 (a) to 8 (c). The level adjustment is applied to the connecting portion of each key phrase, and the effect of fade-in or fade-out is exerted.

Then, the audio level of the obtained audio waveform WAV8 is adjusted so as to be "1/2" to generate an audio waveform WAV9. The audio waveform WAV generated by performing continuity and level adjustment of key phrases of all audio data extracted by the CPU 12 in this manner.
9 is reproduced, and the speaker 21 generates a sound based on the sound waveform WAV9. That is, in the "mode 2", the key phrases of the extracted audio data are repeatedly reproduced one by one.

Then, with respect to the audio data in which the key phrase is reproduced, the color of the icon displayed on the display device 18 changes, and the size of the icon dynamically changes according to the volume. This allows
Recognition of which audio data is being reproduced among the audio data icons displayed on the display device 18 for the key phrase reproduced from the speaker 21 is also visually easy.

FIG. 9 is a diagram showing a change in the icon of the voice data accompanying the reproduction of the key phrase. The case where the key phrase of the voice data in which the voice data recognition word is set to "Wanwan" is reproduced. Is shown. FIG.
(A) shows the icon IC2 when the volume of the key phrase to be reproduced is high, and FIG. 9 (b) shows the icon IC2 when the volume is low. FIG.
The icon IC2 shown in (a) and (b) has a different color compared to other icons, making it easier to identify the icon being reproduced. As described above, since the size of the icon dynamically changes according to the volume of the key phrase emitted from the speaker 21 and the color of the icon also changes, the identification of the audio data being reproduced can be easily visually recognized. Is realized as follows.

Further, in "mode 2" in this embodiment, the voice data recognition word set for the voice data corresponding to the key phrase being reproduced is displayed in a flowing manner. "Flow display" means that characters and the like displayed on the display device 18 move so as to flow on the screen. This example is shown in FIG. FIG. 10 is a diagram showing a flow display of the speech data recognition word in this embodiment. FIG. 10A shows the first stage in which the voice data recognition word is displayed in a flowing manner, and FIG. 10B shows the second stage. FIG. 10C shows the third stage. First, in the first stage, all "lean / lean" set as speech data recognition words are displayed. In the second stage, the voice data recognition word has moved to the left by one character, and the first "L" of "Lean Lean" has disappeared. Further, in the third stage, the character is further moved leftward by one character from the second stage, and the first "Lee" of "Lean Lean" disappears and "L" appears in the right column. Similarly, the voice data recognition word gradually moves to the left,
It is realized that the characters disappear from the left end while the characters appear from the right end. As described above, by displaying the voice data recognition word in a flowing manner, it is easy to visually identify the voice data of the key phrase being reproduced.

As described above, according to the "mode 2", the key phrases of the audio data extracted by the search are sequentially reproduced, and the identification of the reproduced audio data is visually facilitated. As a result, the audio data desired by the operator can be reliably specified from the plurality of extracted audio data. However, if there is much audio data extracted in "mode 2",
In some cases, the time required to reproduce the key phrases of all audio data becomes longer.

A-3) Mode Switching As described above, the mode can be switched between "mode 1" and "mode 2" by selecting an arbitrary mode in the initial setting of the voice database system. It is also possible to switch modes automatically. As a method of automatically switching the mode, if the number of audio data extracted as a result of the search is equal to or more than a predetermined number, the mode is reproduced / displayed in “mode 1”. "Is displayed / displayed. The designated number can be set in advance by the operator.

B) Control such as Display in Temporary Selection State A case where at least one temporarily selected audio data is included in the audio data that matches the search condition when the search result is displayed will be described. In the reproduction in the non-selection state, the volume emitted from the speaker 21 was the level adjusted to “１ ／”. However, in the reproduction in the temporary selection state, the volume generated from the speaker 21 was adjusted to the level. Is output at the volume indicated by the original audio data without performing the above processing. That is, the display and reproduction in the temporary selection state are the same as the contents described in “mode 2” in the non-selection state.
The difference is that no level adjustment is performed.

The key phrases for all of the temporarily selected audio data set by the operation of the operator are reproduced in a state in which they are successively arranged. When a plurality of key phrases are set for one audio data, they are repeated in order. In addition, the effect of fade-in and fade-out is applied to a portion where a key phrase is connected. The color of the icon of the audio data corresponding to the key phrase being reproduced changes, and the size of the icon dynamically changes according to the volume. Further, a voice data recognition word of voice data corresponding to the key phrase being reproduced is displayed in a flowing manner.

As described above, when displayed and reproduced in the tentatively selected state, the key phrases are reproduced sequentially and continuously, and the display of the icon corresponding to the voice data and the display of the voice data recognition word change. Therefore, it is possible to easily recognize the reproduced audio data visually. It is also possible to set a plurality of audio data to be in the temporary selection state.

Further, for example, when the voice data extracted as a result of the search in the non-selection state is displayed / reproduced in “mode 2”, arbitrary voice data is temporarily selected by a mouse operation, so that the operator It is possible to narrow down the desired audio data.

If the operator cannot determine the desired voice data between the non-selection state and the temporary selection state described above, the display device 18 can further display the attribute information. The icon of any audio data on the screen on which the search result is displayed as shown in FIG.
By double-clicking on, the attribute information display screen P4 as shown in FIG. 11 can be displayed on the display device 18 for the audio data. The operator can check the attribute information of the audio data from the attribute information display screen P4 in FIG. By clicking the play button B21 displayed on the attribute information display screen P4 by operating the mouse 20, the entire section of the audio data is reproduced, and the speaker 21 emits sound.
The reproduction in the non-selection state and the provisional selection state is reproduction only in the section of the set key phrase, but when the reproduction button B21 is clicked, all of the audio data is reproduced.

When the operator recognizes that the audio data is the desired audio data by confirming the attribute information and reproducing the audio data, the operator presses the takeout button B22.
By clicking, the audio data can be extracted. "Retrieving audio data" means copying audio data from an audio database so that the audio data can be used by another program or the like. When the operator recognizes that the audio data is not the desired audio data by checking the attribute information or reproducing the audio data, the operator clicks the cancel button B23 to end the attribute information display screen P4 and display the search result. Returning to the screen or the like, the operation of specifying the desired audio data is performed again.

As described above, when the search result of the audio data of the audio database system of this embodiment is displayed, the audio data extracted by the search is automatically reproduced. The voice data can be easily specified as compared with the display of the search result not accompanied by the operator, and the operator does not feel uneasy. Also,
When displaying search results, instead of a file name, an audio data recognition word that specifically and succinctly indicates the type of audio data is displayed in the lower column of the audio data icon. Therefore, it is possible to easily estimate the content of the audio data as compared with the related art, and it is possible to improve the efficiency of the operation. Furthermore, for the reproduction performed when displaying the search results, the key phrase set for the audio data is repeatedly reproduced instead of reproducing all the audio data, so that the desired audio data can be specified in a short time. Becomes possible. In this way, audio data desired by the operator can be obtained promptly and efficiently.

<4. Flowchart> Next, a description will be given of processing from search to identification of voice data in the voice database system of the present embodiment.

FIG. 12 is a flowchart showing processing from search to identification of voice data in the voice database system of this embodiment. First, in step S21, the CPU 12 activates a program related to search / retrieval of voice data among voice database programs in accordance with an instruction of an operator. And
On the initial setting screen, one of mode 1 and mode 2 is selected and set as the reproduction mode in the non-selected state (step S22). Then, in step S23, a condition (for example, a keyword or the like) for searching for voice data is input. Then, the operator inputs a search start instruction in step S24. As a result, the CPU 12 starts searching for audio data that matches the input search condition. The audio data extracted as a result of the search is displayed on the display device 18, and the key phrase of the audio data extracted in the "mode 1" or "mode 2" reproduction mode in the non-selected state is reproduced. (Step S25). Then, the operator determines whether it is necessary to check only specific voice data (step S26). If "YES" is determined here, the process proceeds to a step S27, and if "NO", the process proceeds to a step S29.

In step S27, the operator clicks on the icon of the specific voice data by mouse operation or moves the mouse pointer to change the non-selected state to the temporary selected state. Then, in step S28, display / reproduction in the temporary selection state is performed.

If there is candidate voice data corresponding to the desired voice data, the process proceeds to step S30, and if there is no candidate voice data, the process proceeds to step S36 (step S29). Then, the operator selects the candidate voice data and performs a mouse operation to display an attribute information display screen for the voice data (step S3).
0). Then, final confirmation of the audio data is performed on the attribute information display screen (step S31). And step S3
In Step 2, when all the audio data is reproduced, the process of Step S33 is performed. In step S33, a reproduction instruction is issued by depressing the reproduction button on the attribute information display screen. Then, as a result of the confirmation of the audio data, it is determined whether or not the audio data is taken out (step S3).
4). When the voice data is to be taken out, an operation of taking out is performed (step S35).

If the next search is to be performed, the process from step S22 is repeated. If not, the process proceeds to step S37 to terminate the program related to voice data search / extraction.

Next, the automatic reproduction process in the "mode 1" in the non-selected state will be described. FIG. 13 is a flowchart showing a reproduction process in “mode 1” in a non-selected state of the audio database system according to the present embodiment. First, a list of audio data extracted as a result of the search in step S41 is created and stored in the memory 13. Then, based on the list created in step S41, the reproduction process of the key phrase of the audio data is started for the number of audio data extracted as a result of the search.

For example, assuming that the number of voice data extracted as a result of the search is N (where N is an arbitrary integer), each key phrase of the extracted voice data is expressed as “1 / (2 · N )) And start the playback process. As a result, the key phrases of all the extracted voice data are summed and synthesized, and “1 /
2 ".

The reproduction in "mode 1" is repeated until "NO" is determined in step S43. When the reproduction is completed, the reproduction is stopped by stopping each reproduction process in step S44.

Next, the automatic reproduction process in the "mode 2" in the non-selected state will be described. FIG. 14 is a flowchart showing a reproduction process in “mode 2” in a non-selected state of the audio database system according to the present embodiment. First, a list of audio data extracted as a result of the search in step S51 is created and stored in the memory 13. Then, it is determined whether or not the previously reproduced key phrase has ended (step S52). When the reproduction is being performed, the process proceeds to step S58, and when the reproduction is completed, the process proceeds to step S53. In step S53, each process involved in the reproduction of the key phrase of the audio data whose reproduction has been completed is stopped. In step S54, audio data to be reproduced next is specified based on the list created in step S51. In step S55, step S5
A flow display process is started to display the voice data recognition word of the voice data specified in step 4 in a flow display. In step S56, a reproduction process for reproducing the key phrase of the audio data specified in step S54 is started. In step S57, an icon display process is started to dynamically change the icon display of the audio data specified in step S54. Then, the process proceeds to step S58. Steps S55 to S57 may be performed in another order.

In step S58, it is determined whether or not reproduction in "mode 2" is to be continued. If there is no input by the operator, the process proceeds to step S52, and the reproduction is continued.
When ending the reproduction in “mode 2”, step S5
In step 9, the processes involved in the reproduction of the key phrase of the audio data (the processes started in steps S55 to S57) are stopped, and the process ends.

Next, the automatic reproduction process in the temporary selection state will be described. FIG. 15 is a flowchart showing a reproduction process in a temporarily selected state of the audio database system in this embodiment. First, a list of audio data extracted as a result of the search in step S61 is created and stored in the memory 13. Then, it is determined whether or not the previously reproduced key phrase has ended (step S).
62). If it is during reproduction, the process proceeds to step S68,
If it has been completed, the process proceeds to step S63. In step S63, each process related to the reproduction of the key phrase of the audio data whose reproduction has been completed is stopped. In step S64, the audio data to be reproduced next is specified based on the list created in step S61. Step S
In step 65, a flow display process is started to display the voice data recognition word of the voice data specified in step S64 in a flow display. In step S66, step S6
A reproduction process for reproducing the key phrase of the audio data specified in step 4 is started. In step S67,
The icon display process is activated so that the icon display of the audio data specified in step S64 is dynamically changed. Then, the process proceeds to step S68. Steps S65 to S67 may be performed in another order.

In step S68, it is determined whether or not the reproduction in the provisionally selected state is to be continued. If there is no input by the operator, the process proceeds to step S62, and the reproduction is continued. When the reproduction in the tentatively selected state is ended, each process involved in reproducing the key phrase of the audio data in step S69 (the process started in steps S65 to S67)
Is stopped and the processing is terminated.

<5. Modifications> The icons of the audio data shown in the above-described embodiment are shown by the picture of the speaker. However, the present invention is not limited to this, and the operator can freely set each audio data. For example, if a picture corresponding to the content of audio data is set as an icon,
The visual effect is enhanced, and the audio data can be specified more efficiently.

In the reproduction in the non-selection state, the level is adjusted so that the sound volume becomes "1/2". However, the present invention is not limited to this. Since the temporary selection state is a state where the operator intentionally outputs the sound of the specific sound data, the volume may be increased. However, the reproduction in the non-selected state is automatically performed almost simultaneously with the display of the search result. Therefore, if the volume is large, the operator may feel uncomfortable. Therefore, such a problem is solved by reducing the volume of reproduction in a non-selected state that is not intentional reproduction. Therefore, if the volume is to be reduced, a numerical value other than "1/2" may be used.

[0068]

As described above, according to the first aspect of the present invention, the section voice information based on the start point and the end point of an arbitrary characteristic portion of the voice indicated by the voice data is converted into the voice data. Since the audio data is stored in association with each other, audio data is searched based on predetermined information from a plurality of audio data, and the audio data extracted by the search is reproduced based on the section audio information and output as audio, It is possible to specify voice data quickly, and it is possible to specify desired voice data in a short time without leaving the operator feeling uneasy, and to quickly and efficiently specify voice data desired by the operator. Can be reliably obtained.

According to the second aspect of the present invention, it is possible to store a plurality of section voice information in association with one voice data, so that the voice data can be specified more reliably and easily. be able to.

According to the third aspect of the present invention, the display corresponding to the audio data output by the output means changes along with the reproduction based on the section audio information of the audio data, so that the reproduction is performed. It becomes easy to visually identify the voice data, and voice data desired by the operator can be obtained quickly, efficiently and reliably.

According to the fourth aspect of the present invention, the speech data extracted by the search means is a speech onomatopoeic sound of a characteristic portion or a speech expressed as a character including an objective object of the speech data. Since the data recognition word is displayed, it is easy to estimate what kind of voice each of the voice data extracted as a result of the search is.

According to the fifth aspect of the present invention, the audio data can be easily specified by the computer reading and executing the audio database program recorded on the computer-readable recording medium, Anxiety does not remain for the operator, and desired voice data can be specified in a short time.
It is possible to realize a voice database system that can reliably and quickly obtain voice data desired by an operator.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a configuration of a voice database system according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram for explaining a method of setting a key phrase of audio data according to the embodiment of the present invention.

FIG. 3 is a flowchart showing a process of registering audio data in the audio database system according to the embodiment of the present invention;

FIG. 4 is a diagram showing an example of a display of a search result in the audio database system according to the embodiment of the present invention;

FIG. 5 is a diagram showing a non-selected state and a tentatively selected state of the voice database system according to the embodiment of the present invention;

FIG. 6 is a diagram showing key phrases of two audio data.

FIG. 7 is an explanatory diagram showing the sum total synthesis of key phrases in the voice database system according to the embodiment of the present invention.

FIG. 8 is an explanatory diagram of an audio data reproduction mode in “mode 2” of the audio database system according to the embodiment of the present invention;

FIG. 9 is a diagram showing a change in an icon of audio data accompanying reproduction of a key phrase in the audio database system according to the embodiment of the present invention.

FIG. 10 is a diagram showing a flow display of speech data recognition words in the embodiment of the present invention.

FIG. 11 is a conceptual diagram showing an attribute information display screen according to the embodiment of the present invention.

FIG. 12 is a flowchart showing a process from searching to specifying audio data in the audio database system according to the embodiment of the present invention;

FIG. 13 is a flowchart showing a reproduction process in “mode 1” in a non-selected state of the audio database system according to the embodiment of the present invention.

FIG. 14 is a flowchart showing a reproduction process in “mode 2” in a non-selected state of the audio database system according to the embodiment of the present invention.

FIG. 15 is a flowchart showing a reproduction process in a temporarily selected state of the audio database system according to the embodiment of the present invention.

FIG. 16 is an explanatory diagram showing an operation procedure in a conventional voice database system on a screen displayed on a display device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 I / O device 12 CPU 13 Memory 14 Storage part 15, 16, 17 Interface 18 Display device 19 Keyboard 20 Mouse 21 Speaker D Portable recording medium

Claims (5)

[Claims] 1. A computer for handling audio data, comprising: (a) storage means for storing section audio information of an arbitrary characteristic portion of the audio indicated by the audio data in association with the audio data; A search unit for searching for voice data based on predetermined information from among the voice data, and (c) outputting the voice data extracted by the search unit based on the section voice information and outputting as voice. Means, and a voice database system. 2. The audio database system according to claim 1, wherein said storage means is capable of storing a plurality of said section audio information in association with one audio data. 3. The system according to claim 1, wherein the display corresponding to the audio data output by the output unit changes with the reproduction of the audio data based on the section audio information. Characteristic voice database system. 4. The system according to claim 1, wherein the audio data extracted by the search unit is:
A speech database system for displaying a speech data recognition word expressed by a character including an onomatopoeic word of a pronunciation of a characteristic portion or an objective object of the speech data. 5. A computer, comprising: (a) storage means for storing, in association with the audio data, section audio information on an arbitrary characteristic portion of the audio indicated by the audio data; And (c) an output unit that reproduces the audio data extracted by the search unit based on the section audio information and outputs the audio data as audio. A computer-readable recording medium on which an audio database program is recorded.