JP2000089789A - Voice recognition device and recording medium - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To provide a speech recognition device capable of reliably acquiring a recognition result of input speech. SOLUTION: When an instruction to start speech recognition is input,
The voice recognition management unit 3 causes the voice recognition unit 4 to start a voice recognition process of the voice input by the voice input unit 2, and when an instruction to stop voice recognition is input, refers to the state information holding unit 31 and performs voice recognition. If the unit 4 is in the process of speech recognition, the stop flag 51 is set without giving a stop instruction, and the stop of the recognition process is delayed until the speech recognition unit 4 acquires the recognition result.
Upon acquiring the recognition result, the speech recognition unit 4 outputs the acquired recognition result to the output unit 6 and refers to the stop flag 51. If the flag is set, the speech recognition unit 4 stops the recognition process.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech recognition for starting or stopping a process of recognizing a speech input by a speech input means such as a microphone in response to a speech recognition start or stop instruction given by a user. The present invention relates to an apparatus and a recording medium on which a computer program for speech recognition is recorded.

[0002]

2. Description of the Related Art In recent years, speech recognition application software for recognizing speech input from input means such as a microphone has been developed. By the way, the audio function of a popular personal computer equipped with a microphone is generally used by switching between an audio input channel and an output channel. In this case, it is not possible to input the sound from the microphone while the sound is being reproduced by the speaker. Therefore, in such software, one of the buttons is set as a button for instructing voice recognition, and while the voice is being played, a mechanism is provided in which a user can input voice from the microphone while pressing this button. are doing.

[0003] Further, a high-performance model having an audio function capable of inputting sound from a microphone while sound is being reproduced by a speaker has been developed. When performing voice recognition on such a model, there is an advantage that voice recognition is performed while the voice is being reproduced, without the need to input a voice while pressing the button as described above. However, there is a problem in that voice recognition is performed when voice is input from the microphone even when is not necessary. In addition, since the processing of voice recognition has a large load, there is a risk that the processing speed of other application software will be reduced or the processing will be stopped.
Therefore, even in the speech recognition of such a model, the speech recognition is performed when necessary, and the speech recognition is not performed when not necessary. Is improved.

[0004]

FIG. 1 is a diagram showing a conventional start and stop timing of speech recognition according to a user's instruction to start / stop speech recognition to a speech recognition apparatus. When the user instructs to start voice recognition by pressing a button or the like and speaks into the microphone, the device inputs a voice signal from the microphone and starts voice recognition. When the user instructs to stop speech recognition by pressing a button or the like, the device stops speech recognition.

However, there is an indefinite time lag from the end of inputting the voice to the end of the voice recognition process due to, for example, a difference in the processing speed of the apparatus. Therefore, as shown in FIG. 1, when a stop is instructed during the recognition process and the stop is performed immediately, the recognition process is interrupted and a part that is not recognized remains, and the recognition result cannot be obtained.

SUMMARY OF THE INVENTION The present invention has been made to solve such a problem, and when speech stop processing is instructed when stop of speech recognition is instructed, the stop of the recognition processing is determined. It is an object of the present invention to provide a speech recognition device and a recording medium in which a computer program for speech recognition is recorded, by which the recognition result of the inputted speech can be reliably obtained by delaying until speech is obtained.

[0007]

FIG. 2 shows the principle of the present invention. An instruction to start or stop speech recognition given by the user is input from the instruction input unit 1 to the speech recognition management unit 3. The speech recognition management unit 3 receives status information of the speech recognition unit 4 notified from the speech recognition unit 4 in response to a request from the speech recognition management unit 3 or in response to the start / stop of recognition of the speech recognition unit 4. (Speech recognition stopped, waiting, recognition)
Is provided with a state information holding unit 31 for holding the information. The voice recognition management unit 3 provides a start instruction when a voice recognition start instruction is input from the instruction input unit 1 and a voice recognition unit 4 when the state of the voice recognition unit 4 is not being recognized when a stop instruction is input. An instruction to stop is given to the voice recognition unit 4.

On the other hand, the voice recognition management section 3
When the state information notified from the voice recognition unit 4 when a stop instruction is input from the device indicates that recognition is in progress, the stop flag 51 of the stop information holding unit 5 is set and the stop instruction is issued. Do not give to. When the recognition process ends, the voice recognition unit 4 refers to the stop flag 51, and if the flag instructs the stop, notifies the voice recognition management unit 3 of the stop and stops. As a result, the end of the recognition process is delayed until the recognition result is obtained. When the stop flag 51 is set as described above and the voice recognition start instruction is input from the instruction input unit 1 before the voice recognition unit 4 acquires the voice recognition result, the voice recognition management unit 3 resets the stop flag 51.

Further, the voice recognition management section 3 outputs the state information held by the state information holding section 31 to the output section 6. In addition, the speech recognition management unit 3 performs processing in a result guarantee mode in which the stop of speech recognition processing is delayed until a recognition result is obtained, or stops the recognition processing immediately when stop is instructed. In the case of a speech recognition apparatus that can select which processing is to be performed, the result non-guaranteed flag 32 holds information on which processing mode is selected. The apparatus retains the state of the result non-guaranteed flag 32 even after the speech recognition processing is completed, and reproduces the previous processing mode when the speech recognition processing is restarted.

When an instruction to start speech recognition is given from the speech recognition management section 3 while the recognition is stopped, the speech recognition section 4 takes in a speech signal input from the speech input section 2 and starts speech recognition processing. At the same time, it notifies the voice recognition management unit 3 that recognition is being performed. When the recognition result is obtained, the stop flag 51 of the stop information holding unit 5 is referred to, and the stop flag 51
If is not set, the acquired recognition result is output to the output unit 6 and the voice recognition management unit 3 is notified that the voice recognition is waiting. However, when the stop flag 51 is set, the acquired recognition result is output to the output unit 6, and the stop is notified to the speech recognition management unit 3 to stop the speech recognition processing. At this time, the stop flag 51 is reset.

According to a first aspect of the present invention, there is provided a speech recognition apparatus for starting or stopping speech recognition processing in response to an instruction to start or stop speech recognition. An input unit, a voice input unit for inputting a voice, and an instruction inputting a voice recognition stop instruction by the instruction input unit, until the recognition of the input voice is completed before the input of the instruction is completed. Voice recognition management means for delaying the stop of recognition.

In the first invention, when an instruction to stop speech recognition is inputted by the instruction input means, the speech recognition managing means,
The stop of the voice recognition is delayed until the recognition of the voice input before the instruction is input is completed. Thereby, the recognition result of the input speech can be reliably obtained.

According to a second aspect of the present invention, in addition to the first aspect, the speech recognition apparatus further comprises: first information holding means for holding stop information indicating that an instruction to stop speech recognition has been input by the instruction input means; The voice input by the input unit is recognized to obtain a recognition result. After the recognition result is obtained, the first information holding unit is referred to, and when the stop information is held, the voice recognition process is stopped. Voice recognition means for notifying the voice recognition management means of status information indicating the status each time the state of processing is changed and resetting the processing state; output means for outputting a recognition result obtained by the voice recognition means; Second information holding means for holding state information of the means, wherein the voice recognition management means stores the state information notified from the voice recognition means in the second information holding means, and performs voice recognition by the instruction input means. Start of When the instruction is inputted, the speech recognition means is instructed to start the recognition processing of the speech inputted by the speech input means, and when the instruction input means is inputted, the second information is outputted. Referring to the holding unit, when the voice recognition unit is in a standby state for voice recognition, it instructs the voice recognition unit to stop the voice recognition process, while when the voice recognition unit is performing the voice recognition process, By causing the first information holding unit to hold the stop information, the first information holding unit is a unit that causes the voice recognition unit to continue the voice recognition process at least until a recognition result is obtained.

In the second invention, the speech recognition managing means stores the status information notified from the speech recognizing means in the second information holding means, and when an instruction to start speech recognition is input, the speech recognition managing means. Causes the voice recognition means to start recognition processing of voice input by the voice input means, and when an instruction to stop voice recognition is input, refers to the second information holding means and causes the voice recognition means to wait for voice recognition. If the voice recognition unit is in the middle of the process, a stop instruction is given to the voice recognition unit. Stop information indicating that the recognition has been performed is stored in the first information storage unit, and the stop of the recognition process is delayed until the voice recognition unit obtains a recognition result.
Upon acquiring the recognition result, the voice recognition unit outputs the obtained recognition result to the output unit, and refers to the first information holding unit, and if the stop information is not stored, the voice input unit inputs the next voice. If the stop information is held, the acquired recognition result is output to the output unit, while the recognition processing is stopped to reset the stop information, and the voice recognition management is performed every time the processing state changes. Notify the means of status information. Thereby, the recognition result of the input speech can be reliably obtained.

According to a third aspect of the present invention, in addition to the second aspect of the present invention, the speech recognition apparatus further comprises a second output means for outputting the state information. This makes it possible to clearly indicate to the user whether or not the voice recognition unit is performing the voice recognition process after the instruction to stop the voice recognition is issued.

According to a fourth aspect of the present invention, there is provided the voice recognition apparatus according to the second or third aspect.
In addition to the invention, when the speech recognition is being processed when the instruction to stop the speech recognition is input by the instruction input unit,
Designation means for designating whether the speech recognition means should continue speech recognition processing at least until a recognition result is obtained or immediately stop speech recognition processing, and hold information indicating the content designated by the designation means The voice recognition management means refers to the third information storage means when an instruction to stop voice recognition is input by the instruction input means, and immediately performs voice recognition processing. In a case where it is designated to stop, regardless of whether or not the voice recognition means is performing the voice recognition processing, a means for immediately stopping the voice recognition processing is provided.

In the fourth invention, when the stop of the speech recognition is instructed, the speech recognition processing is continued until the recognition result is obtained when the stop is instructed by referring to the third information holding means. Is specified, the stop of the recognition process is delayed until the recognition result is obtained, while the voice recognition process is stopped immediately when the stop is instructed. Even if the means is performing recognition and the recognition result is not obtained, the voice recognition processing is immediately stopped. Thus, the user can designate a processing mode in which the voice recognition is interrupted and the processing immediately shifts to another processing.

According to a fifth aspect of the present invention, in addition to the fourth aspect, at the end of the voice recognition process, the voice recognition device stores information held by the third information holding device, At the time of activation, referring to the information stored in the means, if speech recognition processing is being performed when an instruction to stop speech recognition is input by the instruction input means, at least until the recognition result is obtained, Means for designating whether the recognition means should continue the speech recognition processing or immediately stop the speech recognition processing. As a result, the same processing mode as the previous one can be reproduced at the time of restart.

According to a sixth aspect of the present invention, there is provided the speech recognition apparatus according to the second or third aspect.
In the voice recognition management means of the invention, when the voice recognition means is in the process of voice recognition and the first information holding means holds the stop information, an instruction to start voice recognition is input by the instruction input means. In this case, there is provided a means for resetting the stop information held by the first information holding means without instructing the speech recognition means to start the speech recognition processing.

In the sixth invention, an instruction to start speech recognition is inputted by the instruction input means when the speech recognition means is in the process of speech recognition and the first information holding means holds the stop information. In this case, the speech recognition management means resets the stop information held by the first information holding means without instructing the speech recognition means to start the speech recognition processing, and thereafter, the speech recognition means which has obtained the recognition result. When the user refers to the first information holding unit, the stop information is not held, and the recognition standby state is established. Thereby, even after the user once instructs the stop of the voice recognition, the voice recognition can be continued as long as the recognition result is not obtained.

A recording medium according to a seventh aspect of the present invention is used in a speech recognition apparatus in which a computer starts or stops speech recognition processing in response to an instruction to start or stop speech recognition, and is a computer-readable recording medium. In the above, a program code means for causing the computer to receive an instruction to start or stop speech recognition, a program code means for causing the computer to accept an input of speech, and an instruction to stop speech recognition to the computer. When input, program code means for delaying the stop of voice recognition until the recognition of voice input before the instruction is input is completed.

In the seventh invention, when the computer reads the computer program on the recording medium and receives an instruction to stop the speech recognition, the computer reads the computer program until the recognition of the speech input before the instruction is completed is completed. Delay stop of speech recognition. Thus, even if the stop of the recognition process is instructed, the computer program of the voice recognition that delays the stop of the recognition process until the recognition result is surely acquired is realized by the CD-R.
It can be provided via a portable recording medium such as an OM or MO, or via a line.

[0023]

FIG. 3 is a diagram showing a flow of an operation in the speech recognition apparatus of the present invention. The speech recognition device of the present invention has the configuration shown in the principle diagram of FIG.
Description of each part is omitted. Immediately after the start-up of the apparatus, the status information “stop” is stored in the status information storage unit 31 of the voice recognition management unit 3. At this time, even if a voice is input from the voice input unit 2, the voice recognition unit 4 Do not recognize. In this state, when an instruction to start speech recognition given by the user is input from the instruction input unit 1 to the speech recognition management unit 3, the speech recognition management unit 3 gives a start instruction to the speech recognition unit 4 and And the status information “waiting for voice recognition” of the voice recognition unit 4 notified from the voice recognition unit 4 in response to the request is stored in the status information storage unit 31.

The speech recognition unit 4 takes in a speech signal input from the speech input unit 2 to start speech recognition, and notifies the speech recognition management unit 3 of "under recognition". If an instruction to stop speech recognition is input during recognition by the speech recognition unit 4, the state information held by the state information holding unit 31 is "recognizing". The stop flag 51 of the unit 5 is set and a stop instruction is not given to the speech recognition unit 4, and the end of the recognition process is delayed until the recognition result is obtained.

When the speech recognition unit 4 acquires the recognition result,
The stop flag 51 of the stop information holding unit 5 is referred to, and since the stop flag 51 is set, the obtained recognition result is output, while the speech recognition management unit 3 determines that “stopped”.
To stop the speech recognition process. At this time, the speech recognition unit 4 resets the stop flag 51.

FIG. 4 is a diagram showing an example of the start and stop timings of speech recognition in the speech recognition apparatus of the present invention. FIG. 4A shows the timing when the stop of speech recognition is instructed during the second recognition after the first recognition result is output. As shown in the figure, even if a stop is instructed during recognition, the stop of the recognition process is delayed and the recognition process is continued until a recognition result is obtained. FIG. 4B shows the timing when an instruction to stop speech recognition is issued during the standby after outputting the first and second recognition results. As shown in the figure, when the stop is instructed during the standby, the recognition process is immediately stopped.

FIG. 5 is a diagram showing another example of the start and stop timings of speech recognition in the speech recognition apparatus of the present invention. This example shows the timing when the stop of the speech recognition is once instructed during the recognition, but the start of the speech recognition is instructed before the recognition result is obtained. As shown in the figure, the stop flag 5
1 is set once, but the start flag is reset before the recognition result is obtained, so the stop flag 51 is reset. Therefore, after the speech recognition unit 4 acquires the recognition result, even if the stop flag 51 is referred to, the stop flag 51 is reset.
When a voice is input from, voice recognition starts. In the above description, the case where the speech recognition unit 4 resets the stop flag 51 when the speech recognition unit 4 stops the recognition processing in response to the stop instruction has been described. However, as shown in FIG. , The stop flag 51 may be reset.

[0028]

FIG. 6 is a block diagram in the case where the speech recognition apparatus of the present invention is a word recognition apparatus. The same or corresponding parts as in the principle diagram of FIG. 2 are denoted by the same reference numerals, and in this example,
The keyboard 1 corresponds to an instruction input unit, the microphone 2 corresponds to a voice input unit, and the word recognition unit 4 corresponds to a voice recognition unit. When the voice recognition device is a word recognition device, the word recognition unit 4 converts the voice signal input from the microphone 2 into
A word is identified by comparing it with a template of a word list 41 as shown in FIG. 7, which stores a word ID and a template (recognized word) of the voice signal.

FIG. 8 is a diagram showing an example of a screen display of the state information and the recognition result of the voice recognition processing, taking the above-described word recognition apparatus as an example. Immediately after the activation of the device, the status information of “stopped” held in the status information holding unit 31 of the voice recognition management unit 3 is displayed.
The word recognizing unit 4 does not perform word recognition even if a voice is input from the. In this state, when the user operates the [S] key (start key) to give an instruction to start recognition, the state information of the word recognition unit 4 changes to “waiting for recognition”.

The word recognizing unit 4 receives a voice signal of "Good morning" input from the microphone 2 and starts voice recognition. At this time, the state information of the word recognition unit 4 changes to “under recognition”. If the user operates the [E] key (stop key) to input an instruction to stop speech recognition while the word recognition unit 4 is recognizing, the state information of the word recognition unit 4 does not change from "recognizing". At this time, the speech recognition management unit 3 sets the stop flag 51 of the stop information holding unit 5 and does not give a stop instruction to the word recognition unit 4, but delays the end of the recognition process until a recognition result is obtained.

When the word recognition unit 4 acquires the recognition result, the word recognition unit 4 refers to the stop flag 51 of the stop information holding unit 5 and outputs the acquired recognition result “Good morning” because the stop flag 51 is set. At this time, the state information of the word recognizing unit 4 changes to “stopped”.

The speech recognition device of the present invention is not limited to a word recognition device, but may be a speaker recognition device for identifying a speaker. The instruction input unit 1 is not limited to a keyboard, but may be a mouse, a soft keyboard, a button icon, or the like. Further, the voice input unit 2 may be a device such as a digital voice recorder, which inputs voice data recorded on a recording medium such as a miniature card by line input, or may be loaded into a computer of a voice recognition device in addition to inputting voice using a microphone. The voice may be input by reading from the recorded recording medium. The output unit 6 is not limited to a display, but may be a printer, a speaker, or the like.

The voice recognition apparatus of the present invention stores the above-described computer program for voice recognition on a magnetic disk,
A portable recording medium such as a CD-ROM, a line destination memory provided in a center, for example, which can communicate a program with a computer wirelessly or by a wire, or a processing device side memory such as a RAM or a hard disk provided in a computer Load from a recording medium such as.

[0034]

As described above, the speech recognition apparatus and the recording medium of the present invention stop the recognition processing when the speech recognition processing is in progress when the stop of the speech recognition is instructed, and obtain the recognition result. As a result, an excellent effect is obtained that the recognition result of the input speech can be reliably obtained.

[Brief description of the drawings]

FIG. 1 is a diagram showing start and stop timings of conventional speech recognition.

FIG. 2 is a principle diagram of the present invention.

FIG. 3 is a diagram showing a flow of operation in the voice recognition device of the present invention.

FIG. 4 is a diagram illustrating an example of start and stop timings of speech recognition.

FIG. 5 is a diagram showing another example of the start and stop timings of speech recognition.

FIG. 6 is a specific block diagram of the speech recognition device of the present invention.

FIG. 7 is a conceptual diagram of a word list.

FIG. 8 is a diagram illustrating a screen display example of state information and a recognition result of a voice recognition process.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Instruction input part (keyboard) 2 Speech input part (microphone) 3 Speech recognition management part 31 State information holding part 32 Result unguaranteed flag 4 Speech recognition part 41 Word list 5 Stop information holding part 51 Stop flag 6 Output part (display)

Claims (7)

[Claims] 1. A speech recognition apparatus for starting or stopping speech recognition processing in response to an instruction to start or stop speech recognition, comprising: an instruction input means for inputting an instruction for starting or stopping speech recognition; A voice input means for performing the voice recognition, and a voice for delaying the stop of the voice recognition until the recognition of the voice input before the input of the voice recognition instruction is completed by the instruction input means. A speech recognition device comprising: a recognition management unit. 2. A first information holding means for holding stop information indicating that an instruction to stop speech recognition has been input by an instruction input means, and a recognition result obtained by recognizing a voice input by the voice input means. After obtaining the recognition result and referring to the first information holding means, if the stop information is held, the voice recognition processing is stopped to reset the stop information, and each time the state of the processing changes, Voice recognition means for notifying state information representing the state to the voice recognition management means, output means for outputting a recognition result obtained by the voice recognition means, and second information holding means for holding state information of the voice recognition means; The voice recognition management means stores the state information notified from the voice recognition means in the second information holding means, and when a voice recognition start instruction is input by the instruction input means, the voice input means By The speech recognition means instructs the speech recognition means to start recognition processing of the inputted speech, and refers to the second information holding means when the instruction input means inputs an instruction to stop speech recognition. When the recognition is waiting, the voice recognition means is instructed to stop the voice recognition processing. On the other hand, when the voice recognition means is performing the voice recognition processing, the stop information is held in the first information holding means. 2. The speech recognition apparatus according to claim 1, wherein the speech recognition unit causes the speech recognition unit to continue the speech recognition processing at least until a recognition result is obtained. 3. The speech recognition device according to claim 2, further comprising a second output unit that outputs the status information. 4. When voice recognition processing is being performed when an instruction to stop voice recognition is input by the instruction input means,
Designation means for designating whether the speech recognition means should continue speech recognition processing at least until a recognition result is obtained or immediately stop speech recognition processing, and hold information indicating the content designated by the designation means The voice recognition management means refers to the third information storage means when an instruction to stop voice recognition is input by the instruction input means, and immediately performs voice recognition processing. 4. The voice according to claim 2, further comprising means for immediately stopping the speech recognition processing when the speech recognition means is designated to be stopped, regardless of whether or not the speech recognition means is performing the speech recognition processing. Recognition device. 5. A means for storing the information held by the third information holding means at the end of the voice recognition processing, and referring to the information stored at the next activation of the voice recognition processing. If the speech recognition processing is being performed when the instruction for stopping speech recognition is input by the instruction input means, the speech recognition means may continue the speech recognition processing at least until a recognition result is obtained, or 5. The speech recognition apparatus according to claim 4, further comprising means for designating whether to stop the speech recognition processing. 6. The voice recognition managing means instructs the start of voice recognition by the instruction input means when the voice recognition means is in the process of voice recognition and the first information holding means holds stop information. 4. The apparatus according to claim 2, further comprising: means for resetting the stop information held by the first information holding means without instructing the speech recognition means to start the speech recognition processing when the input is input. Voice recognition device. 7. A computer-readable recording medium that is used in a speech recognition device in which a computer starts or stops speech recognition processing in response to an instruction to start or stop speech recognition, wherein: Program code means for receiving an input of an instruction to start or stop voice recognition; program code means for causing the computer to receive an input of voice; and when an instruction to stop voice recognition is input to the computer, A program code means for delaying the stop of voice recognition until the recognition of voice input before the instruction is input is completed.