JP6966374B2 - Speech recognition system and computer program - Google Patents

Description

The present invention relates to a voice recognition technique for recognizing a user's spoken voice.

As a voice recognition technology for recognizing a user's spoken voice, for each word registered in the voice recognition dictionary in advance, the likelihood of expressing the plausibility that the word is the word represented by the spoken voice is calculated, and the plausibility is calculated. A technique is known in which the maximum word is recognized as a word spoken by a user only when the likelihood exceeds a predetermined threshold value.

Further, in such a voice recognition technology, in the voice recognition performed for the voice of the previous user, the recognition fails because there is no word whose likelihood exceeds the threshold value, and the voice of the user this time. If the word recognized in the voice recognition performed for the previous user is the word with the highest likelihood in the voice recognition performed for the utterance voice of the previous user, the word recognized this time is used in the previous voice recognition. There is known a technique of determining a false rejection (FA) and changing the threshold value to a value at which the word is more easily recognized (for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2007-41319

According to the above-mentioned technology for changing the threshold value, when the user speaks the same word in the previous speech recognition and the current speech recognition, the word is recognized as the word with the maximum likelihood in the previous speech recognition. However, if the word is recognized by this voice recognition, the threshold value cannot be changed even though the word recognized this time by the previous voice recognition is erroneously rejected, and the efficiency is high. False rejection rate (FRR) cannot be reduced.

Further, according to the above-mentioned technique for changing the threshold value, if the word having the maximum likelihood in the previous speech recognition is the word recognized in the previous speech recognition, the threshold value is changed. However, if the maximum likelihood of the word in the previous speech recognition is low, it is possible that the user has not spoken the same word as the previous one, that is, the previous speech recognition was falsely rejected (FA; False Rejection). ), Therefore, the change of the threshold value tends to cause a false acceptance (FA) in which the wrong word is recognized as the word.

Therefore, it is an object of the present invention to efficiently reduce false rejection rate (FRR) without increasing false acceptance rate (FAR) as much as possible in speech recognition.

In order to achieve the above object, the present invention has a voice recognition system in which words are recognized by voice, a microphone, a voice recognition dictionary in which a plurality of words in which reference values indicating the degree of matching are set are registered, and the voice. A word registered in the recognition dictionary that is higher than the degree represented by the preliminary reference value set to represent the degree of matching lower than the reference value set for the word in the voice picked up by the microphone. The word that matches in is detected as a preliminary recognition word, and the degree of the word registered in the voice recognition dictionary that is higher than the degree represented by the reference value set in the word in the voice picked up by the microphone. When the voice recognition means that outputs the words matching in the above as the recognition result and the voice recognition means detect the preliminary recognition word, the preliminary recognition count value of the word detected as the preliminary recognition word is incremented by 1. After that, when the predetermined period elapses, the preliminary recognition counting means that decrements the preliminary recognition counting value of the word by 1, and the word that is output as the recognition result when the voice recognition means outputs the recognition result. When the preliminary recognition count value of is greater than or equal to a predetermined value (however, the predetermined value is an integer of 2 or more), the reference value of the word output as the recognition result is changed to indicate a lower degree of matching. It is equipped with a means for changing the reference value.

Here, in such a voice recognition system, when the reference value changing means changes the reference value of the word output as a recognition result so as to represent a lower degree of matching, the spare of each word. It may be configured to clear the recognition count value.

Further, in order to achieve the above-mentioned problems, the present invention comprises a voice recognition system in which words are recognized by voice, a microphone, and a voice recognition dictionary in which a plurality of words in which reference values indicating the degree of matching are set are registered. A word registered in the speech recognition dictionary, which is greater than or equal to the degree represented by a preliminary reference value set to represent a degree of matching lower than the reference value set for the word in the voice picked up by the microphone. A word that matches with a high degree is detected as a preliminary recognition word, and a word registered in the voice recognition dictionary that is more than the degree represented by the reference value set in the word in the voice picked up by the microphone. When the voice recognition means that outputs a word that matches to a high degree as a recognition result and the voice recognition means detects the preliminary recognition word, the preliminary recognition count value of the word detected as the preliminary recognition word is set to 1. When the number increases and then a predetermined period elapses, the preliminary recognition counting means that decreases the preliminary recognition counting value of the word by 1 and the voice recognition means output the recognition result, the recognition result is output. When the preliminary recognition count value of the word is equal to or greater than a predetermined value (however, the predetermined value is an integer of 2 or more), the reference value of the word output as the recognition result indicates a lower degree of matching. It is provided with a reference value change proposal means for proposing such a change, and a reference value editing means for changing the reference value of the word according to a user's operation.

Here, such a voice recognition system proposes to change the reference value of the word output as a recognition result so as to represent a lower degree of matching in the reference value change proposal means. It may be configured to clear the preliminary recognition count value of each word.

Further, in each of the above voice recognition systems, the predetermined value may be an integer of 3 or more.
Further, the above voice recognition system may be a voice recognition system used for voice input in an information processing system mounted on an automobile.
In the above voice recognition system, in the voice recognition system, only when a voice similar to the voice that utters the recognized word is input multiple times in the period immediately before the recognition of the word, the reference of the word is given. Change the value to a value that represents the degree of matching, or propose the change. Here, in the situation immediately before the recognition of such a word, a situation in which a voice similar to the voice in which the recognized word is uttered is input multiple times is a situation in which the user repeatedly utters until the same word is recognized. That is, it is highly probable that the word can be correctly recognized in response to the reoccurrence of the word in which False Rejection (FR) has occurred. In the situation where the voice similar to the voice uttering the recognized word is input three or more times in the period immediately before the recognition of the word, the voice similar to the voice uttering the recognized word is input three times or more. Since it is the input situation, the probability is particularly high.

Therefore, according to the speech recognition system as described above, only for a word in which a true false rejection (FR) has occurred, the reference value is changed to indicate a lower degree of matching, and the word is changed. It can be easily recognized. Therefore, the false rejection rate (FRR) can be reduced without increasing the false acceptance rate (FAR).

In the voice recognition system as described above, whether or not the voice input in the period immediately before the recognition of the word is similar to the voice that utters the recognized word is determined by the voice as a backup of the word. Since it is determined whether or not the words are matched to a higher degree than the reference value indicates, even if the most matched word is not the recognized word, the voice input in the period immediately before the recognition of the word is erroneously rejected ( False Rejection; FR) can be detected to change the reference value to a value indicating the lower degree of matching, or to propose the change. On the other hand, even if the most consistent word input in the period immediately before the recognition of the word is the recognized word, if the consistency is low, it is similar to the voice uttering the recognized word. Since it is not detected as voice, it is possible to prevent false rejection (FR) from being detected and a change in the reference value or a proposal for the change.

As described above, according to the present invention, in speech recognition, the false rejection rate (FRR) can be efficiently reduced without increasing the false acceptance rate (FAR) as much as possible. Can be done.

It is a block diagram which shows the structure of the information processing system which concerns on embodiment of this invention. It is a figure which shows the voice recognition dictionary and the threshold value table which concerns on embodiment of this invention. It is a figure which shows the voice recognition method of the voice recognition engine which concerns on embodiment of this invention. It is a flowchart which shows the FR correspondence threshold value adjustment processing which concerns on embodiment of this invention. It is a figure which shows the processing example of the FR correspondence threshold value adjustment processing which concerns on embodiment of this invention. It is a figure which shows the threshold value adjustment screen which concerns on embodiment of this invention. ..

Hereinafter, an embodiment of the present invention will be described by taking as an example an application to an information processing system mounted on an automobile.
As shown in the figure, the information processing system includes a data processing unit 1, a microphone 2, a voice input unit 3, an input device 4, a display device 5, and other peripheral devices 6 such as a camera, an audio device, and a GPS receiver. ..

Here, the voice input unit 3 recognizes the voice of the user input from the microphone 2 and outputs the recognition result to the data processing unit 1.
The data processing unit 1 is provided with various functions such as a car navigation function, a music player function, and a camera-captured image display function, and performs processing according to the recognition result input from the voice input unit 3.

Next, the voice input unit 3 also includes a voice recognition engine 31, a voice recognition dictionary 32, a threshold table 33, and a voice input control unit 34.
Here, such an information processing system may be configured by using a computer including a CPU, a memory, a peripheral device, and the like, and in this case, the above-mentioned data processing unit 1 and voice input unit 3 may be configured. It may be realized by the CPU executing a computer program.

Next, as shown in FIG. 2a, a plurality of words to be recognized by the voice recognition engine 31 and an identification number (No.) of the words are registered in the voice recognition dictionary 32.
Further, in the threshold value table 33, for each word to be recognized by the voice recognition engine 31, the identification number (No.) of the word, the threshold value Th of the word, and the threshold value of the word are displayed. The presence or absence of adjustment indicating whether or not to perform the adjustment of is registered.

Next, the operation of voice recognition performed by the voice recognition engine 31 will be described.
The voice recognition engine 31 calculates the score of each word stored in the voice recognition dictionary 32 for the recognition target voice in parallel with the input of the recognition target voice which is the voice input from the microphone 2.

Here, the score of each word registered in the speech recognition dictionary 32 for the recognition target voice represents a predicted value of the magnitude of the difference between the word and the phrase represented by the recognition target voice, and predicts a larger difference. The more you do, the higher your score.

More specifically, in the calculation of the score, after setting a predetermined initial value as the score, each time the sound of each voice section of the recognition target voice (for example, the voice section for each phonetic element) is input, the score is calculated. Whether or not there is a match between the sound of the voice section and the pronunciation of the part corresponding to the voice section of each word registered in the voice recognition dictionary 32 is calculated, and if they match, the score is reduced by a predetermined value and the match is made. If not, the score is increased by a predetermined value. The increase / decrease value of the score for each voice section of the voice to be recognized is, for example, the value obtained by multiplying the ratio of the words in the voice section to the total voice section by the initial value of the score.

According to such voice recognition, when the recognition target voice is "aiueoka", the transition of the score calculated for the word "aiueo" is shown in FIG. 3a, and it is calculated for the word "aiueoka". As shown in FIG. 3b, the score with the word gradually decreases and the sound of the recognition target voice that does not match the word is input while the sound of the recognition target voice that matches the word is input. During this time, the ward score will gradually increase.

That is, for example, as shown in FIG. 3a, the recognition target voice "aiueoka", the word "aiueo", and the score gradually decrease during the period in which the recognition target voice "aiueo" sound is input, and then gradually decrease. , Increases when the "ka" of the recognition target voice is input.

Similarly, as shown in FIG. 3b, the recognition target voice "aiueoka", the word "aiai", and the score gradually decrease during the period in which the recognition target voice "ai" sound is input. After that, the period during which the recognition target voice "Eoka" is input gradually increases.

By the way, in the voice recognition engine 31, the score of the recognition target voice calculated as described above and any word is equal to or less than the threshold value Th registered in the threshold value table 33 of the word. If so, the word whose score is equal to or less than the threshold value Th is output to the voice input control unit 34 as a recognition word, and the voice input control unit 34 processes the recognition word output from the voice recognition engine 31 as a recognition result. Output to part 1.

That is, for example, in the case of the word "aiueo" shown in FIG. 3a, the score for the word "aiueo" becomes equal to or less than the threshold value Th when the "e" of the recognition target voice "aiueoka" is input. At this point, the recognition word "aiueo" is output.

On the other hand, in the case of the word "aiai" shown in FIG. 3b, the score for the word "aiai" is never equal to or less than the threshold value Th, so this word "aiai" is not output as a recognition word. ..

Further, if the voice recognition engine 31 has a score of the recognition target voice calculated as described above and one of the words equal to or less than the preliminary recognition threshold Pth set for the word, the voice recognition engine 31 determines. It also performs a process of pre-recognizing a word whose score is equal to or less than the pre-recognition threshold value Pth and outputting it to the voice input control unit 34 as a pre-recognition word.

Here, the value below the preliminary recognition threshold Pth of each word is the value obtained by adding a predetermined value to the threshold Th registered in the threshold table 33 of that word, or the threshold table 33 of that word. It is assumed that the threshold value Th registered in is increased by a predetermined ratio of the threshold value Th.

By setting the preliminary recognition threshold value Pth in this way, for example, in the case of the word "aiueo" shown in FIG. 3a, the "e" of the recognition target voice "aiueoka" is input and the score is high. Before the recognition word "aiueo" is output when the value is Th or less, the score becomes less than or equal to the preliminary recognition threshold PTh when the "u" of the recognition target voice "aiueoka" is input, and the word "aiueo" "Aiueo" is pre-recognized and the pre-recognition word "aiueo" is output.

On the other hand, the word "aiuei" shown in FIG. 3b is not output as a recognition word because the score does not fall below the threshold value Th, but the "u" of the recognition target voice "aiueoka" is input. At that point, the score becomes equal to or less than the preliminary recognition threshold PTh, the word "aiai" is pre-recognized, and the preliminary recognition word "aiai" is output. Next, the voice input control unit 34 receives a false rejection rate (FRR; The FR-compatible threshold adjustment process performed to reduce False Rejection Rate) will be described.

FIG. 4 shows the procedure of this FR-compatible threshold adjustment process.
As shown in the figure, the voice input control unit 34 generates a preliminary recognition word output from the voice recognition engine 31 (step 402) and a timer timeout described later (step 402) in the FR-compatible threshold adjustment process. 404), the generation of the output of the recognition word from the speech recognition engine 31 and (step 406) are monitored.

Then, if the output of the preliminary recognition word occurs during the monitoring of steps 402, 404, and 406, it is registered in the threshold table 33 of the preliminary recognition word (the word that is the preliminary recognition word). It is checked whether or not the adjustment is made (step 412), and if the adjustment is not made, the process returns to the monitoring of steps 402, 404, and 406.

The initial value of the presence / absence of adjustment registered in the threshold table 33 is adjusted for all words.
On the other hand, when the presence / absence of adjustment registered in the threshold table 33 of the pre-recognized word is adjusted (step 412), a flag is set for the pre-recognized word (step 412). Step 414), the timer associated with the set flag is started (step 416), and the process returns to the monitoring of steps 402, 404, and 406. Here, in the timer started in step 416, a predetermined time (for example, 10 seconds) is set as the timeout time. However, the timeout time of the timer may be set so that the longer the word, the larger the timeout time length, according to the length (number of characters) of the pre-recognized words.

Next, if a timer timeout occurs during the monitoring of steps 402, 404, and 406 (step 404), the flag associated with the timer for which the timeout has occurred is cleared (step 422). .. Then, the process returns to the monitoring of steps 402, 404, and 406.

Next, if the recognition word is output from the voice recognition engine 31 during the monitoring of steps 402, 404, and 406 (step 406), the recognized word (word that is the recognition word) is word. It is checked whether or not the adjustment is made in the threshold table 33 of the above (step 432), and if the adjustment is not made, the process returns to the monitoring of steps 402, 404, and 406.

On the other hand, when the presence / absence of adjustment registered in the threshold table 33 of the recognized word is adjusted (step 432), the number of flags set for the recognized word is It is checked whether or not the predetermined value n (n is, for example, 3) or more (step 434), and if it is not the predetermined value n or more, the process returns to the monitoring of steps 402, 404, and 406.

On the other hand, if the number of flags set for the recognized word is a predetermined value n or more (step 434), the threshold value Th registered in the threshold table 33 for the recognized word is set. , Increase by a predetermined value (step 436).

Then, all the flags of each word set at the present time are cleared (step 438), and the process returns to the monitoring of steps 402, 404, and 406.
The FR-compatible threshold adjustment process performed by the voice input control unit 34 has been described above.
In the above FR-compatible threshold adjustment process, the flag of each word functions as a counter for the word whose number of flags represents the count value, and the above FR-compatible threshold adjustment process is performed in step 414. The counter of the pre-recognized word is incremented by 1, step 416 is the process of starting the timer associated with the pre-recognized word, and step 422 is the process associated with the timer in which the timeout has occurred. Step 438 is a process of clearing the counters of all words, and step 434 is performed assuming that the counter value of the counter of the recognized word represents the number of flags of the word. Even so, it is equivalent.

When the FR-compatible threshold adjustment process is performed using a counter in this way, in the following description, the counter value of the counter of each word is treated as representing the number of flags of the word.

Here, FIG. 5 shows a processing example of such an FR-compatible threshold adjustment process.
In the illustrated example, no word is recognized even if the user says "Chizukakudai", so as a result of repeating saying "Chizukakudai" again, the user's third "Chizukakudai" It is about the case where the word "Chizukakudai" is recognized for the utterance.

That is, it is a case where the word can be correctly recognized for the third utterance representing the word after the false rejection (FR) is repeated twice for the same word. ..

Further, in this example, it is assumed that the timeout time of the timer used for the FR-corresponding threshold adjustment process shown in FIG. 4 is 10 seconds, and the predetermined value n is 3.
In this case, as shown in the figure, the voice spoken by the user "Chizukakudai" is repeatedly input from the microphone 2 to the voice recognition engine 31 as the recognition target voice three times.
In this case, the voice recognition engine 31 sets the threshold value Th of the first recognition target voice "Chizukakudai" and the second recognition target voice "Chizukakudai" as the score of either word. The following scores are not calculated and the recognition word is not output. Then, after that, the voice recognition engine 31 first, for the third recognition target voice "Chizukakudai", with respect to the word "Chizukakudai" among the words registered in the voice recognition dictionary 32. The score that is less than or equal to the word threshold Th is calculated, and the recognition word "Chizukakudai" is output (t4).

On the other hand, since the voice uttering the word "Chizukakudai" is at least similar to the recognition target voice "Chizukakudai" each time, the voice recognition engine 31 says the first recognition target voice "Chizukakudai" 2 For the second recognition target voice "Chizukakudai", the score that is equal to or less than the preliminary recognition threshold PTh of the word "Chizukakudai" is calculated as the score of the word "Chizukakudai", and the preliminary recognition word "Chizukakudai" Is output (t1, t2). Also, for the third recognition target voice "Chizukakudai", the preliminary recognition threshold value of the word "Chizukakudai" is used as the score of the word "Chizukakudai" before the recognition word "Chizukakudai" is output. Calculate the score below PTh and output the preliminary recognition word "Chizukakudai" (t3). Then, each time the voice input control unit 34 outputs the preliminary recognition word "Chizukakudai", the voice input control unit 34 sets a flag for the word "Chizukakudai" and maintains the set state for 10 seconds after setting.

In addition, the voice recognition engine 31 has a predetermined level or higher and similar utterances to the recognition target voice "Chizukakudai" other than the word "Chizukakudai" for each recognition target voice "Chizukakudai". For words as well, a score that is equal to or less than the preliminary recognition threshold PTh of the other word is calculated as the score of the other word, the other word is output as the preliminary recognition word, and the voice input control unit 34 is concerned. Whenever another word is output as a preliminary recognition word, the flag for the other word is set, and after setting, the set state is maintained for 10 seconds.

Then, if the recognition word "Chizukakudai" is output at time t4, the voice input control unit 34 sets the number of flags set for the word "Chizukakudai" output as the recognition word. In the example investigated and illustrated, since it is 3 or more, which is a predetermined value n, the threshold value Th of the word “chizukakudai” is increased. However, it is assumed that the time from the output of the first preliminary recognition word "Chizukakudai" at time t1 to the output of the recognition word "Chizukakudai" at time t4 is within 10 seconds of the timeout time.

In such FR-compatible threshold adjustment processing, a plurality of recognition target voices "Chizukakudai" including the word "Chizukakudai" are included for the first recognition target voice "Chizukakudai" and the second recognition target voice "Chizukakudai". Is output as a preliminary recognition word, and even if the score of the word "Chizukakudai" is not the minimum among the scores of the words output as the preliminary recognition word, the recognition word "Chizukakudai" When is recognized, the threshold Th of the word "Chizukakudai" is increased. That is, according to the FR-corresponding threshold adjustment process as described above, when there are a plurality of words output as preliminary recognition words for both the first recognition target voice and the second recognition target voice, the word is specified. The threshold value Th of the word output as the recognition word is increased for the third recognition target voice among the plurality of words.

Then, by increasing the threshold value Th of the word "Chizukakudai", the word "Chizukakudai" becomes more easily recognized, and thereafter, the false rejection rate (FRR) of the word "Chizukakudai" becomes easier to recognize. Rate) decreases.

For example, if the threshold value Th of the word "Chizukakudai" is increased to the value of X in Fig. 4, the same recognition target voice as the first and second recognition target voice "Chizukakudai" in Fig. 4 On the other hand, the word "Chizukakudai" can be recognized as a recognition word.

Here, the voice in which the same word as the recognized word is detected as the preliminary recognition word is a voice similar to the voice in which the recognition word is uttered.
Then, as described above, in the FR-compatible threshold adjustment process, in the period immediately before the recognition of the recognition word, a voice similar to the voice uttering the recognition word (voice in which the preliminary recognition word is detected) is produced. The threshold value Th of the word is increased only when it is input multiple times. In addition, in the period immediately before the recognition of such a recognized word, the situation in which a voice similar to the voice uttering the recognized word is input multiple times is a situation in which the user repeatedly utters until the same word is recognized. That is, it is highly probable that the word can be correctly recognized as a recognition word for the reoccurrence of the word in which the false rejection (FR) has occurred. In the situation where the voice similar to the voice uttering the recognition word is input three or more times in the period immediately before the recognition of the recognition word, the voice similar to the voice uttering the recognition word is input three times or more. Since this is the situation, the probability is particularly high.

Therefore, it is possible to increase the threshold value Th only for words for which false rejection (FR) has truly occurred to make the word easier to recognize, and false acceptance rate (FAR). False rejection rate (FRR) can be reduced without increasing the false rejection rate (FRR).

It should be noted that, as described above, whether or not the voice input in the period immediately before the recognition of the recognition word is similar to the voice in which the recognized recognition word is spoken is determined with respect to the voice as a preliminary for the recognition word. Since it is determined by whether or not a score equal to or lower than the recognition threshold PTh is calculated, even if the word with the best matching voice input in the period immediately before the recognition of the recognition word is not the same word as the recognition word. False rejection (FR) can be detected and the threshold Th of the word can be increased. On the other hand, if the word with the best matching voice input in the period immediately before the recognition of the recognition word is the same word as the recognition word, but the score does not fall below the preliminary recognition threshold PTh. Since the voice is not detected as a voice similar to the voice spoken by the recognition word, it is suppressed that the false rejection (FR) is mistakenly detected and the threshold Th of the word is increased. NS.

Here, in the above-mentioned FR-corresponding threshold adjustment process, the threshold Th registered in the threshold table 33 is increased by a predetermined value with respect to the word recognized as the recognition word in step 436. However, step 436 displays a message prompting the threshold of the word recognized as a recognized word, for example, "" Chizukakudai "becomes easier to recognize when the threshold is increased". Alternatively, it may be a process of outputting voice. However, in this case, the data processing unit 1 is provided with a function of changing the threshold value Th of each word of the threshold value table 33 according to the user operation, and the user can use the process. Allow yourself to adjust the word threshold.

Alternatively, the data processing unit 1 displays the threshold value adjustment screen for each word on the display device 5 as shown in FIG. 6a, and the threshold value table is displayed according to the user operation for the threshold value adjustment screen. A threshold value editing function for changing the threshold value Th of 33 is provided so that the user can adjust the threshold value of the word by himself / herself, and step 436 is performed in the threshold value editing function. The threshold value change process of the word recognized as the recognition word may be started by the attribute of the threshold value increase proposal type. Here, when the threshold value editing function activates the threshold value change process of a specific word, the threshold value adjustment screen for the word shown in FIG. 6a is displayed on the display device 5, and the threshold value adjustment is performed. While accepting the user operations of the increase key 601 and the decrease key 602 provided on the screen as threshold adjustment operations as shown in FIGS. 6b and 6c, the threshold value Th of the word in the threshold value table 33 is changed. In addition, when the threshold value editing function activates the threshold value change process with the threshold value increase proposal type attribute, "If the setting is increased," map enlargement "will be more easily recognized on the threshold value adjustment screen. Include a message that suggests reducing the threshold, such as "Masu".

By the way, in addition to the above processing, the voice input control unit 34 determines that the threshold value is decreased according to the user operation for the word whose threshold value Th is increased by the FR-compatible threshold value adjustment processing. It also performs a process of setting whether or not to adjust the threshold table 33 of the word without adjustment.

The embodiment of the present invention has been described above.

1 ... Data processing unit, 2 ... Microphone, 3 ... Voice input unit, 4 ... Input device, 5 ... Display device, 6 ... Peripheral device, 31 ... Voice recognition engine, 32 ... Voice recognition dictionary, 33 ... Value table, 34 ... Voice input control unit.

Claims (8)

It is a voice recognition system that recognizes words by voice.
With a microphone
A speech recognition dictionary in which multiple words with reference values indicating the degree of matching are registered, and
A word registered in the voice recognition dictionary, which is greater than or equal to the degree represented by a preliminary reference value set to represent a degree of matching lower than the reference value set for the word in the voice picked up by the microphone. A word that matches with a high degree is detected as a preliminary recognition word, and a word registered in the voice recognition dictionary that is more than the degree represented by the reference value set in the word in the voice picked up by the microphone. A voice recognition means that outputs words that match to a high degree as a recognition result,
When the voice recognition means detects the preliminary recognition word, the preliminary recognition count value of the word detected as the preliminary recognition word is incremented by 1, and if a predetermined period elapses thereafter, the preliminary recognition of the word is performed. Pre-recognition counting means that reduces the counting value by 1 and
When the voice recognition means outputs the recognition result, the preliminary recognition count value of the word output as the recognition result is a predetermined value or more (however, the predetermined value is an integer of 2 or more). A voice recognition system comprising a reference value changing means for changing the reference value of a word output as a recognition result so as to represent a lower degree of matching. The voice recognition system according to claim 1.
The reference value changing means is characterized in that when the reference value of a word output as a recognition result is changed to represent a lower degree of matching, the preliminary recognition count value of each word is cleared. Speech recognition system. It is a voice recognition system that recognizes words by voice.
With a microphone
A speech recognition dictionary in which multiple words with reference values indicating the degree of matching are registered, and
A word registered in the voice recognition dictionary, which is greater than or equal to the degree represented by a preliminary reference value set to represent a degree of matching lower than the reference value set for the word in the voice picked up by the microphone. A word that matches with a high degree is detected as a preliminary recognition word, and a word registered in the voice recognition dictionary that is more than the degree represented by the reference value set in the word in the voice picked up by the microphone. A voice recognition means that outputs words that match to a high degree as a recognition result,
When the voice recognition means detects the preliminary recognition word, the preliminary recognition count value of the word detected as the preliminary recognition word is incremented by 1, and if a predetermined period elapses thereafter, the preliminary recognition of the word is performed. Pre-recognition counting means that reduces the counting value by 1 and
When the voice recognition means outputs the recognition result, when the preliminary recognition count value of the word output as the recognition result is a predetermined value or more (however, the predetermined value is an integer of 2 or more), the user. With respect to the reference value change proposal means that proposes to change the reference value of the word output as the recognition result so as to indicate a lower degree of matching.
A voice recognition system comprising a reference value editing means for changing the reference value of the word according to a user's operation. The voice recognition system according to claim 3.
The reference value change proposing means clears the preliminary recognition count value of each word when proposing to change the reference value of the word output as a recognition result so as to represent a lower degree of matching. A voice recognition system characterized by that. The voice recognition system according to claim 1, 2, 3 or 4.
A voice recognition system characterized in that the predetermined value is an integer of 3 or more. The voice recognition system according to claim 1, 2, 3, 4 or 5.
The voice recognition system is a voice recognition system characterized by being a voice recognition system used for voice input in an information processing system mounted on an automobile. A computer program that is read and executed by a computer equipped with a microphone.
The computer program is the computer,
A speech recognition dictionary in which multiple words with reference values indicating the degree of matching are registered, and
A word registered in the voice recognition dictionary, which is greater than or equal to the degree represented by a preliminary reference value set to represent a degree of matching lower than the reference value set for the word in the voice picked up by the microphone. A word that matches with a high degree is detected as a preliminary recognition word, and a word registered in the voice recognition dictionary that is more than the degree represented by the reference value set in the word in the voice picked up by the microphone. A voice recognition means that outputs words that match to a high degree as a recognition result,
When the voice recognition means detects the preliminary recognition word, the preliminary recognition count value of the word detected as the preliminary recognition word is incremented by 1, and if a predetermined period elapses thereafter, the preliminary recognition of the word is performed. Pre-recognition counting means that reduces the counting value by 1 and
When the voice recognition means outputs the recognition result, the preliminary recognition count value of the word output as the recognition result is a predetermined value or more (however, the predetermined value is an integer of 2 or more). A computer program characterized by functioning as a reference value changing means for changing the reference value of a word output as a recognition result so as to indicate a lower degree of matching. A computer program that is read and executed by a computer equipped with a microphone.
The computer program is the computer,
A speech recognition dictionary in which multiple words with reference values indicating the degree of matching are registered, and
A word registered in the voice recognition dictionary, which is greater than or equal to the degree represented by a preliminary reference value set to represent a degree of matching lower than the reference value set for the word in the voice picked up by the microphone. A word that matches with a high degree is detected as a preliminary recognition word, and a word registered in the voice recognition dictionary that is more than the degree represented by the reference value set in the word in the voice picked up by the microphone. A voice recognition means that outputs words that match to a high degree as a recognition result,
When the voice recognition means detects the preliminary recognition word, the preliminary recognition count value of the word detected as the preliminary recognition word is incremented by 1, and if a predetermined period elapses thereafter, the preliminary recognition of the word is performed. Pre-recognition counting means that reduces the counting value by 1 and
When the voice recognition means outputs the recognition result, the preliminary recognition count value of the word output as the recognition result is a predetermined value or more (however, the predetermined value is an integer of 2 or more). A reference value change proposal means that proposes to change the reference value of the word output as a recognition result so as to indicate a lower degree of matching, and
A computer program characterized by functioning as a reference value editing means for changing the reference value of the word according to a user's operation.

Images