JP2000259181A - Device and method for recognizing speech information, and recording medium where program for recognizing speech information is recorded - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a speech information recognition device, etc., capable of targeting a speech-recognizable part for speech recognition even in the case that speech data differing in compression ratio co-exist in one speech file. SOLUTION: This is a speech information recognition device configured of a personal computer 4 comprising a speech recognition program 9 converting speech information stored in a speech file in predetermined block units into character information by speech-recognizing each block, a control program 8 which checks a compression ratio of the speech information in each block and judges it to be suitable for speech recognition when the compression ratio is smaller than a predetermined one, judges it not to be suitable for speech recognition when the compression ratio is larger the predetermined one, and is made to execute speech recognition and its conversion into character information only when the compression ratio is judged to be suitable for speech recognition, and a display 5 for displaying unsuitableness and giving a warning when the compression ratio is judged to be unsuitable for speech recognition.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice information recognition device, a voice information recognition method, a recording medium on which a program for recognizing voice information is recorded, and more specifically, voice information is recognized and converted into character information. The present invention relates to a voice information recognition device, a voice information recognition method, and a recording medium storing a program for recognizing voice information.

[0002]

2. Description of the Related Art In recent years, digital recorders have been commercialized that convert voice into digital data and record it as a voice file on a flash memory card or the like.

[0003] Such a digital recorder performs encoding in order to reduce the amount of information when recording an audio file.
Some models are provided with an SP (standard play) mode having a relatively low compression ratio and an LP (long play) mode having a relatively high compression ratio. Providing a plurality of recording modes in this way has the advantage that the SP mode can be used when sound quality is emphasized, and the LP mode is used when sound recording time is emphasized. are doing.

[0004] An audio file is recorded in units of blocks each having a size of, for example, 512 bytes, and the recording mode can be changed for each block.

On the other hand, techniques for recognizing speech and converting it to character information have been studied in the past, but in recent years, it has been commercialized as software for information devices such as personal computers, and is in the stage of practical use. Has reached.

When such a speech recognition technique is applied to speech data in a speech file recorded by a digital recorder as described above, in order to avoid a decrease in the recognition rate, the speech recognition of the target speech data is performed. It is necessary to limit the quality to a certain level or more, and in the above-described recording mode, only the sound recorded in the SP mode is subjected to speech recognition, and the sound recorded in the LP mode is subjected to speech recognition. Exclusions have been made.

That is, in the conventional speech recognition technology, a speech file recorded only in the SP mode is targeted for speech recognition, and other speech files are excluded from speech recognition. The unit was an audio file.

[0008]

As described above, the recording in the SP mode and the recording in the LP mode can be performed in units of blocks constituting an audio file. It is conceivable that the recording mode is switched from the SP mode to the LP mode on the way, judging that the remaining capacity has become small.

In such a case, a block recorded in the SP mode and a block recorded in the LP mode are mixed in one audio file.
In the above-described conventional technology, such a voice file is excluded from voice recognition. Therefore, even if it is a part, it cannot respond to a user who wants to perform speech recognition.

The present invention has been made in view of the above circumstances, and even when voice data having different compression ratios are mixed in one voice file, a voice recognizable portion may be subjected to voice recognition. Voice information recognition device,
An object of the present invention is to provide a voice information recognition method and a recording medium on which a program for recognizing voice information is recorded.

[0011]

In order to achieve the above object, a speech information recognition apparatus according to a first aspect of the present invention includes a conversion means for recognizing speech information and converting it to character information, And control means for controlling whether or not the converted audio information is converted into character information in accordance with the compression ratio.

Further, the speech information recognition device according to the second invention is the speech information recognition device according to the first invention,
When the compression rate of the speech information is at a level suitable for speech recognition, the control means causes the conversion means to perform speech recognition to convert the speech information into character information, and the compression rate of the speech information is suitable for speech recognition. When the level is not present, a predetermined warning signal is output.

Further, a voice information recognition device according to a third aspect of the present invention controls the voice information recognition device according to the first or second aspect of the present invention to determine whether or not each voice block is converted into character information. .

A speech information recognition method according to a fourth aspect of the present invention controls whether speech information encoded at a predetermined compression rate is converted into character information in accordance with the compression rate, and converts the information into character information. When the control is performed, the voice information is recognized and converted into character information.

According to a fifth aspect of the present invention, a recording medium storing a program for recognizing voice information is a recording medium storing a program for recognizing voice information by a computer. If it is controlled to convert audio information encoded at a predetermined compression rate into character information according to the compression rate, and to convert the information to character information, the audio information is recognized and the character It is converted into information.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. 1 to 7 show an embodiment of the present invention, and FIG. 1 is a diagram showing an overall configuration of a dictation system.

As shown in FIG. 1, the dictation system converts sound into an electric signal and converts it into sound data, which is recorded in a low compression rate SP (standard play) mode or a high compression rate LP (long play) mode. Digital recorder 1 and the digital recorder 1
A miniature card 2 which is used by being detachably mounted on the card and is a recording medium for recording the audio data;
A PC for inserting the miniature card 2 into a PC card slot 10 (see FIG. 2) described later to enable connection.
A card adapter 3, a display 5 as a warning means, a keyboard 6 for inputting, a mouse 7 and the like are provided, and a control program 8 and a voice recognition program are added to voice data obtained from the miniature card 2 through the PC card slot 10. And a personal computer 4 as a voice information recognition device for performing the processing by the computer 9.

FIG. 2 is a block diagram showing an electrical configuration of the personal computer 4. As shown in FIG.

The personal computer 4 reproduces sound and displays information according to the control program 8.
In addition to performing processing for recognizing voice information and converting it to character information in accordance with the voice recognition program 9, various processing is performed in accordance with other various programs. CPU that doubles as
11, a main memory 12 as a recording medium serving as a work area for the CPU 11, an internal recording medium 13 such as a hard disk or the like in which the control program 8 and the voice recognition program 9 are recorded, and various external devices. And an interface (hereinafter abbreviated as IF) for connecting the display 5 to the external port 14 for connecting to the display 5.
15 and an IF1 for connecting the keyboard 6 and the mouse 7
6 and a speaker 18 for emitting sound based on the sound data
And an IF 17 for connecting the speaker 18 and the PC
It comprises a PC card slot 10 into which the miniature card 2 attached to the card adapter 3 is inserted, and an IF 19 for connecting the PC card slot 10 to the CPU 11, the main memory 12, and the internal recording. Medium 13, external port 14, IFs 15, 16, 1
7, 19 are connected to each other via a bus.

The audio data may be directly read from the miniature card 2 via the PC card slot 10. However, the audio data is temporarily recorded on the internal recording medium 13 and read from the internal recording medium 13. Alternatively, the information may be directly read from the digital recorder 1 via communication means such as infrared, serial, or USB.

The control program 8 and the voice recognition program 9 may be recorded in the internal recording medium 13 before shipment, but in the case of a general-purpose personal computer, the floppy disk on which these programs are recorded is stored. The recording medium such as a disk, CD-ROM, or DVD-ROM may be installed in the internal recording medium 13. At this time, these floppy disks, C
A D-ROM, a DVD-ROM, or the like constitutes a recording medium on which a program for recognizing audio information is recorded.

FIG. 3 is a block diagram showing the structure of an audio file in the dictation system as described above.

This audio file has a file structure of, for example, DSS type.

That is, as shown in FIG. 3A, the audio file is configured in units of blocks each having a size of, for example, 512 bytes, and the first two blocks, for example, are used as a file header. , And the rest are blocks in which audio data is recorded.

The file header includes, for example, the number of block headers, a self-recognition flag, a system version number,
System release number, license ID code, user ID code, job number code, purpose code,
Process management code, status management code, recording start date and time, recording end date and time, recording time, erroneous erasure prevention flag, priority level, destination ID code, multiple I mark addresses, spare area, management code area, file text basic area, A file text spare area and the like are provided.

The audio data has one or more DSS blocks, which are audio information blocks. Each of these DSS blocks has, as shown in FIG.
A DSS block header is provided at the beginning, and audio data is recorded in other parts in frame units.

The DSS block header includes the DSS
Information for determining whether the audio data recorded in the block is recorded at a low compression rate in the SP (standard play) mode or is recorded at a high compression rate in the LP (long play) mode. Is recorded.

That is, this audio file can be switched between recording in the SP mode and recording in the LP mode in units of blocks. As a result, the type of the audio file includes a DSS block composed of only the SP mode, a DSS block composed of only the LP mode, and a DSS block composed of the SP mode and a DSS block composed of the LP mode. There are three types.

FIG. 4 is a flowchart showing the operation of the personal computer as described above when performing voice recognition.

When the operation is started, an audio file recorded on the miniature card 2 is captured via the PC card slot 10 (step S1). And
The data of each file recorded in the management area of the file system and the file header is read (step S2), and the file name information included in the read file data is displayed on the display 5 as a list (step S3). ).

At this time, for example, a display as shown in FIG. 6 is performed on the screen of the display 5. FIG. 6 is a diagram illustrating an example of a screen display when performing voice recognition in a personal computer.

FIG. 6 shows the main screen 21.
A menu bar 22 for selecting an operation related to a file or an operation related to editing, and a tool button bar 2 for displaying various operations in an easily understandable manner using icons.
3, an audio file list box 24 for displaying various data relating to the audio file transferred from the miniature card 2, and a control button 25 for performing processing such as play, stop, fast forward and fast reverse, respectively. The displayed tool button bar 23 is provided with a voice recognition button 26 for starting voice recognition processing.

In the audio file list box 24, a column 31 for displaying a list of file names, a column 32 for displaying a recording time, a column 33 for displaying a recording date and time,
In order to confirm whether or not voice recognition can be performed, a field 34 for displaying a recording mode and an end mark field 35 are provided in order, and based on the file data read in step S2, The information is displayed in S3.

Then, the process waits for the user to select a file to be subjected to the voice recognition process and to give an instruction to perform voice recognition on the selected file (step S4).

Here, the selection of the audio file is performed according to FIG.
As shown in the figure, by clicking on the file portion to be selected in the audio file list box 24 with the mouse 7 or the like, the highlighted portion is displayed.
The selection can be visually confirmed.

When the voice recognition button 26 is clicked with the mouse 7 or the like while a file is selected, an instruction input for voice recognition is made.

If it is confirmed in step S4 that a voice file has been selected and a voice recognition instruction has been given, then the voice file is added to the SP file.
It is determined whether or not the file has a mixture of the recording part in the mode and the recording part in the LP mode (step S5).

Here, if the file is not a mixed file, S
It is determined whether the file is a voice file recorded only in the P mode (step S6). If the file is a voice file in the SP mode, all voice data in the voice file is subjected to voice recognition, and voice recognition is performed. Perform processing (step S
8). When the voice recognition has been completed, the process returns to step S4 to wait for the next file to be selected.

If it is determined in step S6 that the audio file is not an audio file recorded only in the SP mode, it is determined that the audio file is an audio file recorded only in the LP mode. There is no voice data to be subjected to voice recognition in the file. Accordingly, the fact that it is impossible to execute the voice recognition is displayed on the display 5 (step S7), and thereafter, the process returns to the step S4 to wait for another file to be selected.

On the other hand, if it is determined in step S5 that the file is a file in which the recorded portions in the SP mode and the LP mode are mixed, a warning message indicating that there is a part where voice recognition cannot be performed is displayed. Is performed on the display 5 (step S9), and just in case, the user is asked whether to cancel the process (step S10).

Here, if the cancellation of the process is selected, the process returns to the step S4. If the process is selected, the process proceeds to the step S8, and the process proceeds to the step S8. The voice recognition processing is performed only on the part recorded in the SP mode.

Next, FIG. 5 is a flowchart showing details of the voice recognition processing in step S8.

When this operation is started, the DSS block header information of the DSS block to be subjected to voice recognition is read (step S21), and the recording mode (compression mode) of the DSS block is determined (step S22).

Here, when it is determined that the compression mode is the SP mode, speech recognition is executed for the DSS block (step S23), and the result of the speech recognition is output, for example, as shown in FIG. Are displayed on the display 5 as characters (step S24).

FIG. 7 is a diagram showing an example of a display screen of software capable of displaying a sentence, and shows a state in which a result of voice recognition is displayed. This FIG.
In FIG. 7, the portions indicated by reference numerals 42 and 44 in the screen 41 are portions for displaying the recognition results.

When the voice recognition for the DSS block is completed, it is determined whether or not the end of the voice file has been reached (step S26). If not, the process returns to step S21 to return to the next step S21. Perform processing for the DSS block.

If it is determined in step S22 that the compression mode is the LP mode, a speech signal is output to the effect that the speech recognition is not executed for the DSS block. Is displayed on the display 5 as indicated by reference numeral 43 in FIG. 7 (step S24).

In the example shown in FIG. 7, the fact that blocks for which speech recognition cannot be performed for the recording time of 18 seconds is displayed as characters, but the present invention is not limited to this. For example, "?" Symbols such as “*” may be inserted, and the number of symbols to be inserted may be made to be proportional to the time when recognition becomes impossible.

Thus, in step S26,
When it is determined that the end of the voice file has been reached, the voice recognition processing ends.

In the above description, the speech information recognition device is
Although the personal computer that executes the control program and the voice recognition program has been described as an example, the present invention is not limited to this, and may be another general-purpose computer or may be configured as a dedicated voice information recognition device. I do not care.

According to such an embodiment, since it is confirmed in block units (in audio block units) whether or not the compression ratio is suitable for speech recognition, audio recognition is performed in one audio file. Even if there are blocks that are suitable for and some blocks that are not suitable, it is possible to perform voice recognition on the part that can be recognized and convert it into character information.

Further, when the voice recognition is not performed, the fact is displayed on the display, so that the recognition can be easily performed. If the display is performed in proportion to the length of time during which the speech recognition is not performed, the amount can be more intuitively grasped.

It should be noted that the present invention is not limited to the above-described embodiment, and it is needless to say that various modifications and applications can be made without departing from the gist of the invention.

[Appendix] According to the embodiment of the present invention described in detail above, the following configuration can be obtained.

(1) Speech recognition means for recognizing speech information stored in a speech file in units of predetermined blocks and converting it into character information for each block, and whether the speech information is suitable for speech recognition And a control unit for executing the operation by the voice recognition unit only when it is determined that the voice recognition unit is suitable for voice recognition.

(2) The control means determines that the speech information is suitable for speech recognition when the compression rate of the audio information in block units is equal to or less than a predetermined compression rate, and when the compression rate is larger than the predetermined compression rate. The speech information recognition device according to supplementary note (1), which determines that the speech information is not suitable for speech recognition.

(3) If the control means determines that the speech information is not suitable for speech recognition, a warning means for warning the fact is further provided. apparatus.

(4) A step of checking for each block whether or not the voice information stored in the voice file in a predetermined block unit is suitable for voice recognition, and it is determined that the voice information is suitable for voice recognition. And performing speech recognition of the speech information on a block-by-block basis and converting the speech information into character information only in the above case.

(5) A recording medium in which a program for recognizing audio information by a computer is recorded. The program stores, in the computer, audio information stored in a sound file in a predetermined block unit. It is necessary to confirm whether or not it is suitable for speech recognition for each block, and to convert the speech information to text information by speech recognition for each block only when it is determined that the speech information is suitable for speech recognition. A recording medium on which a program for recognizing audio information is recorded.

(6) a recording medium for recording audio information;
Converting means for converting the audio information recorded on the recording medium into character information, and processing the audio information by the converting means;
Control means for controlling the voice information in accordance with the compression ratio of the voice information.

(7) A warning means is further provided, wherein the control means causes the conversion means to perform speech recognition when the compression ratio is within a range in which speech recognition is possible, and outputs character information to enable speech recognition. The voice information recognition apparatus according to (6), wherein when the value is out of the range, the warning means is warned to that effect.

Therefore, according to the invention described in the appendix (1), whether or not the voice information is suitable for voice recognition is confirmed in units of blocks, so that one voice file is suitable for voice recognition. Even in the case where there are voice information blocks that are present and voice information blocks that are not suitable, it is possible to perform voice recognition on the part that can be recognized and convert it into character information.

According to the invention described in Appendix (2),
The present invention has the same effect as the invention described in Appendix (1), and can determine whether or not it is suitable for speech recognition based on the compression ratio.

Further, according to the invention described in the supplementary note (3), the same effect as that of the invention described in the supplementary note (1) is obtained, and a warning is issued by the warning means when the voice recognition is not performed. It is possible to recognize that.

According to the invention described in Appendix (4), whether or not the speech information is suitable for speech recognition is confirmed in block units, so that a speech file suitable for speech recognition is contained in one speech file. Even when the information block and the unsuitable voice information block are mixed, the voice recognizable portion can be converted into character information by performing voice recognition.

According to the invention described in the supplementary note (5), by controlling the computer by the program, it is confirmed whether or not the speech information is suitable for speech recognition in units of blocks. Even if the voice information block suitable for voice recognition and the voice information block not suitable for voice recognition coexist, the voice recognizable portion can be converted to character information by performing voice recognition.

According to the invention described in the appendix (6), since the conversion from the audio information to the character information is controlled in accordance with the compression ratio, audio data having different compression ratios can be stored in one audio file. Even in a mixed case, a portion that can be recognized by speech can be subjected to speech recognition.

According to the invention described in the supplementary note (7), the same effect as the invention described in the supplementary note (6) can be obtained, and a warning is issued by the warning means when the voice recognition is not performed. Can be recognized.

[0069]

As described above, according to the speech information recognition apparatus of the present invention according to the first aspect, since the conversion from speech information to character information is controlled according to the compression ratio, one speech is recognized. Even when voice data with different compression ratios are mixed in a file, it is possible to perform a voice recognition on a part that can be recognized.

According to the second aspect of the present invention, the same effects as those of the first aspect can be obtained, and a predetermined warning signal is output when no voice recognition is performed. This makes it possible to recognize that fact.

Further, according to the voice information recognition apparatus of the present invention, the same effects as those of the first or second aspect can be obtained.

According to the voice information recognition method of the present invention, since the conversion from voice information to character information is controlled according to the compression ratio, different compression ratios in one voice file can be obtained. Even in the case where voice data is mixed, a portion that can be recognized by voice can be subjected to voice recognition.

According to the recording medium storing the program for recognizing speech information of the present invention according to claim 5, by controlling the computer with the program, the conversion from the speech information to the character information can be performed at the compression ratio. Therefore, even when voice data having different compression ratios are mixed in one voice file, a voice recognizable portion can be subjected to voice recognition.

[Brief description of the drawings]

FIG. 1 is a diagram showing an overall configuration of a dictation system according to an embodiment of the present invention.

FIG. 2 is an exemplary block diagram showing an electric configuration of a personal computer in the dictation system of the embodiment.

FIG. 3 is a block diagram showing a structure of an audio file in the dictation system of the embodiment.

FIG. 4 is a flowchart showing an operation when performing voice recognition in the personal computer of the embodiment.

FIG. 5 is a flowchart showing details of a voice recognition process in FIG. 4;

FIG. 6 is an exemplary view showing an example of a screen display when performing voice recognition in the personal computer of the embodiment.

FIG. 7 is an exemplary view showing an example of a display screen showing a result of voice recognition in the personal computer of the embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Digital recorder 2 ... Miniature card 4 ... Personal computer (voice information recognition device) 5 ... Display (warning means) 6 ... Keyboard 7 ... Mouse 8 ... Control program 9 ... Voice recognition program 11 ... CPU (conversion means, control means, Voice recognition means) 12 main memory 13 internal recording medium (recording medium storing a program for recognizing voice information)

Claims (5)

[Claims] 1. A conversion means for recognizing voice information and converting it into character information, and a control for controlling whether or not voice information encoded at a predetermined compression rate is converted into text information according to the compression rate. Means, and a voice information recognition device, characterized by comprising: And a control unit that, when the compression ratio of the speech information is at a level suitable for speech recognition, causes the conversion unit to perform speech recognition to convert the speech information into character information. 2. The speech information recognition apparatus according to claim 1, wherein a predetermined warning signal is output when is at a level not suitable for speech recognition. 3. The voice information recognition device according to claim 1, wherein control is performed to determine whether or not to convert to character information for each voice block. 4. Controlling whether or not voice information encoded at a predetermined compression rate is converted into character information according to the compression rate, and recognizes voice information when control is performed to convert the information into character information. A voice information recognizing method characterized by converting the voice information into character information. 5. A recording medium on which a program for recognizing audio information by a computer is recorded, said program causing the computer to convert audio information encoded at a predetermined compression rate in accordance with the compression rate. A program for recognizing voice information characterized by recognizing voice information and converting it to text information is controlled when the control to convert or not to text information is performed. The recording medium on which it was recorded.