JP3415500B2 - Speech recognition system for mobile phone - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a mobile phone device.
In particular, the present invention relates to a voice recognition system and method for mobile phone devices.

[0002]

2. Description of the Related Art As a conventional technique, JP-A-61-1142
Some are described in Japanese Patent Publication No. 99. This publication describes an example of a method of improving the recognition rate of voice recognition. In the voice recognition, if the similarity value of the first candidate is smaller than a predetermined first threshold value, the second candidate is the second largest. The difference between the similarity of 1 and the similarity of the first candidate is calculated, and when the difference is smaller than the second threshold value, it is determined that the target is not a recognition target, thereby reducing false recognition.

[0003]

However, when voice recognition is used in a noisy environment, the similarity value of the first candidate becomes large, even though the recognition result of the first candidate is not correct. There is a case where the condition of becoming smaller than the threshold value of 1 is not satisfied, and there is a problem that the determination process for reducing the false recognition cannot be performed.

Therefore, in view of the above problems, the present invention makes it possible to reduce erroneous recognition when a voice is input to a voice recognition device in which a large number of similar recognition words are registered in a noisy environment. an object of the present invention is to provide a speech recognition system of telephone equipment.

[0005]

According to the present invention, in order to solve the above problems, a plurality of recognition words indicating a voice and a plurality of noise words are registered in a voice recognition system of a portable telephone device for inputting voice. With a voice recognition unit that holds a dictionary, recognizes the voice input, and outputs recognition results of a plurality of candidates, an appearance ratio of the recognition word candidates recognized by the voice recognition unit, and an appearance ratio of the noise word candidates. , The recognition word or the noise word is determined , the first candidate is the recognition word, and
If the noise word appearance rate is greater than or equal to the judgment value, the recognition result
Is determined as a noise word, and the noise words after the second candidate are output.
If the current ratio is less than the judgment value, the recognition result is judged as a recognition word.
Provided is a voice recognition system for a mobile phone device, which is provided with a recognition determination unit that determines the voice recognition.

By this means, a large number of similar recognition words
Voice input to a registered voice recognition device in a noisy environment.
If it occurs, it is possible to reduce erroneous recognition. That is,
Not only the first candidate of the knowledge result but also the second candidate to the kth candidate
By using multiple recognition results, input during voice recognition
Perform the process of determining whether it is voice or noise, and
The complementary word is the recognition word, and the appearance ratio of the noise words after the second candidate is determined.
If it is more than the fixed value, the recognition result is judged as a noise word.
As a result, accidental misrecognition can be prevented and accurate voice
Enables recognition, prevents erroneous recognition, and provides accurate voice recognition
Became possible.

Further, the present invention is a portable telephone for voice input.
In the speech recognition system of the talking device,
Holds a dictionary in which recognition words and multiple noise words are registered, and
Recognize voice input and output recognition results of multiple candidates
Voice recognition unit and the recognition word recognized by the voice recognition unit
Complementary appearance rate, noise word candidate appearance rate
Or, it is judged whether it is a noise word, and the first candidate is a noise word.
If the appearance ratio of the recognition word after the second candidate is greater than or equal to the judgment value,
In this case, the recognition result is determined to be a recognition word, and the second and subsequent candidates are determined.
If the occurrence ratio of the recognition word of is less than the judgment value, the recognition result
Is provided as a noise word and a recognition determination unit is provided.
A voice recognition system for a mobile phone device is provided.

By this means, as in the above invention, a large number of
Noisy voice recognition devices that have registered similar recognition words
It is possible to reduce erroneous recognition when voice is input in the environment.
Become. That is, not only the first candidate of the recognition result but also the second candidate
By using multiple recognition results from complement to k-th candidate
Determine if the input during speech recognition is voice or noise
Process, the first candidate is a noise word, and the second candidate and later are recognized.
If the linguistic appearance rate is greater than or equal to the judgment value, the recognition result is confirmed.
Since it is determined to be a literary word, accidental misrecognition is prevented.
Yes, accurate voice recognition is possible.

Preferably, the appearance ratio of the noise word,
Appearance ratio of recognition words is multiple noise words, multiple recognition words
To which the noise word and the recognition word are
It is calculated by adding each weight each time it appears. This
By this means, the recognition performance of voice recognition can be improved.

Further, the present invention is a portable telephone for voice input.
In the speech recognition system of the talking device,
Holds a dictionary in which recognition words and multiple noise words are registered, and
Recognize voice input and output recognition results of multiple candidates
Voice recognition unit and the recognition word recognized by the voice recognition unit
Complementary appearance rate, noise word candidate appearance rate
Or, it is determined whether it is a noise word,
Of the above, the recognition consisting of less than a predetermined number of characters
A recognition determination unit that uses word candidates as the noise word candidates.
Voice recognition system for mobile phone devices
I will provide a. By this means, the recognition word
When the recognition word contains a small number of characters,
There is a high possibility that the input at the time of recognition is noise and the number of characters is long.
If so, it can be determined that the possibility of being voice is high. like this
In addition to determining whether the recognition result candidate is voice or noise,
When a recognition word is included in the recognition result candidates, the character
By adding the number to the judgment value, more accurate speech recognition results can be obtained.
It is possible to output the result.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a voice recognition system for a mobile phone device according to the present invention. As shown in the figure, the wireless unit 1 is included in the mobile phone device.
The wireless unit 1 transmits and receives wireless signals to and from a base station (not shown).

Further, the portable telephone device is provided with an operating portion 2, and the operating portion 2 operates the portable telephone, starts voice recognition and the like. Further, the mobile phone device is provided with the display unit 3,
The display unit 3 displays numbers, letters and the like. The wireless unit 1, the operation unit 2, and the display unit 3 have a main CPU (Central Pr).
processing unit (central processing unit) 4 is connected, the main CPU 4 controls transmission / reception of the wireless unit 1, receives a control instruction from the operation unit 2, and performs display control on the display unit 3.

A voice recognition unit 5 is connected to the main CPU 4, and the voice recognition unit 5 is a voice recognition LSI (Large S).
Cale Integrated Circuit (large-scale integrated circuit), and performs voice recognition processing. When the voice recognition start key is pressed by the operation unit 2, the main CPU 4 detects that the voice recognition start key is pressed to confirm the start of voice recognition, and the voice recognition start command is transmitted to the voice recognition unit 5. It

The main CPU 4 and the voice recognition unit 5 have an A / D
(Analog to Digital; analog
The (digital) converter 6 and the voice synthesis unit 10 are connected, and the voice synthesis unit 10 synthesizes the recognition result from the voice recognition unit 5 into voice. The voice synthesizer 10 has a D / A (Digital
to Analog (digital / analog) converter 7 is connected.

A microphone 8 is connected to the A / D converter 6, and the microphone 8 inputs a voice. A / D
The converter 6 converts a voice analog signal into digital voice data. The voice input to the microphone 8 is processed as a transmission sound of the mobile phone.

Further, in the voice recognition section 5, the main CPU
When the start command is received from 4, the recognition process is performed on the voice input from the microphone 8. D / A
A speaker 9 is connected to the converter 7, and the speaker 9 outputs a voice. The D / A converter 7 converts digital voice data into a voice analog signal.

The voice output to the speaker 9 is processed as a reception sound of the mobile phone. Further, when the voice recognition unit 5 sounds the start sound at the start of voice recognition and the recognition result is confirmed by the voice recognition unit 5, the voice corresponding to the recognition result is synthesized by the voice synthesis unit 10 and D / A
It is output from the speaker 9 via the converter 7. Further, the above recognition result determined by the voice recognition unit 5 is displayed on the display unit 3.

Next, the voice recognition section 5 receives a recognition word as
Memory dialing of mobile phone devices, function calls
A plurality of recognition words and a plurality of n noise words that are registered to prevent malfunction due to noise are registered in the dictionary. What is a noise word? When noise is input during voice recognition, a memory dial, a voice recognition dictionary that is registered to prevent malfunction of the recognition word for function calls, relates to multiple types of sounds such as sudden noise. Noise is registered.

By this registration, even if sudden noise is input during voice recognition, when the pattern matching with the voice recognition dictionary is performed, the noise word is the input sound more than the recognition word of the memory dial or function call. Is more similar to.

Therefore, the recognition result becomes a noise word, and it becomes possible to prevent malfunctions due to noise input.
Further, the main CPU 4 is provided with a recognition determination unit 4A, and the recognition determination unit 4A performs a determination process of a recognition word and a noise word which are the recognition results of the voice recognition unit 5.

FIG. 2 is a flow chart for explaining the outline of the operations of the main CPU 4, the voice recognition section 5, and the recognition determination section 4A of FIG. In step S1, the voice recognition start key is pressed from the operation unit 2. In step S2, the main CPU 4 detects the depression of the voice recognition start key.
In step S3, the main CPU 4 finalizes the activation of voice recognition.

In step S4, the main CPU 4 sends a voice recognition start command to the voice recognition unit 5. In step S5, the voice recognition unit 5 starts voice recognition. In step S6, the voice recognition unit 5 performs voice recognition on the voice input from the microphone 8.

In step S7, the voice recognition unit 5 determines the result of voice recognition. In step S8, the recognition results of the first candidate to the kth candidate are output to the recognition determination unit 4A as the recognition result. In step S9, the recognition determination unit 4A adds the second candidate to the kth candidate in addition to the first candidate of the recognition result.
The determination process up to the candidate is performed.

In step S10, the main CPU 4
When the recognition determination unit 4A determines that the recognition result is the recognition word, the maintenance corresponding to the recognition word is displayed on the display unit 3,
The corresponding voice is output from the speaker 9. Step S1
In No. 1, when the recognition determination unit 4A of the main CPU 4 determines that the recognition result is a noise word, it is determined that noise is input during voice recognition and a malfunction occurs. In step S12, a message notifying that the determined noise has malfunctioned is displayed on the display unit 3, and a corresponding voice is output from the speaker 9.

FIGS. 3 and 4 are flow charts for explaining the detailed process of determining the recognition result in the recognition determining unit 4A of the main CPU 4 of FIG. As shown in FIG. 3, when the voice recognition start key is pressed from the operation unit 2 in step S21, the recognition determination unit 4A activates voice recognition, and the voice recognition unit 5 recognizes the voice input from the microphone 8. Then, the recognition process is performed to wait for the recognition result to be confirmed.

In step S22, when the recognition result is confirmed and k candidates from the first candidate to the k-th candidate are output as the recognition result by the voice recognition, the recognition determining section 4A recognizes the k candidates. The results are acquired, and the judgment is performed using them as follows. In step S23, it is first determined whether the first candidate is a recognition word. In step S24, when the first candidate is a recognition word, it is determined whether the second candidate is a noise word. If it is not a noise word, the process proceeds to step S26.

In step S25, when the second candidate is a noise word, weighting w1 [k] is added to the determination value. In step S26, it is next determined whether the third candidate is a noise word. If it is not a noise word, step S
Proceed to 28. In step S27, when the third candidate is a noise word, weighting w1 [k-1] is added to the determination value.

In step S28, it is continuously determined whether each candidate is a noise word up to the kth candidate. If it is not a noise word, the process proceeds to step S30. In step S29, when the k-th candidate is a noise word, weighting w1 [1] is added to the determination value. In step S30, when the determination processing up to the kth candidate is completed, the cumulative addition value of the determination values is compared with the threshold Th1.

In step S31, the judgment value is Th1.
When it is larger than the above, even if the first candidate of the recognition result is a recognition word, the voice input during voice recognition is determined to be noise. In step S32, the recognition result for the noise input is displayed and the process ends. In step S33, when the determination value is smaller than Th1 in step S30, it is determined that the input during voice recognition is voice.

In step S34, the recognition result for the first candidate recognition word is displayed, and the process ends. Therefore, even if erroneous recognition occurs in which the first candidate is a recognition word despite the noise being input during the voice recognition in the voice recognition, the second candidate is determined by using the results of the second candidate to the kth candidate. -When the k-th candidate includes a large number of noise words and the determination value is Th1 or more, it is determined that the input during voice recognition is not a voice but a noise input.

Therefore, it is possible to avoid erroneous recognition by displaying the recognition result for the noise input. As shown in FIG. 4, in step S35, when it is determined in step S23 that the first candidate is a noise word, a process of determining whether or not the second candidate is a recognition word is performed. If it is not a recognition word, the process proceeds to step S37.

In step S36, when the k-th candidate is a recognized word, weighting w2 [k] is added to the determination value.
In step S37, next, it is determined whether the third candidate is a recognition word. If it is not a recognition word, the process proceeds to step S39. In step S38, when the third candidate is a recognition word, weighting w2 [k-1] is added to the determination value.

In step S39, it is continuously determined whether or not each candidate is a recognized word up to the kth candidate. If it is not a recognition word, the process proceeds to step S41. In step S40, when the k-th candidate is a recognition word, the determination value is weighted w2.
Add [1]. In step S41, when the determination process up to the kth candidate is completed, the cumulative addition value of the determination values is compared with the threshold Th2.

In step S42, the judgment value is Th2.
If it is larger than, even if the first candidate of the recognition result is a noise word, it is determined that the voice input during voice recognition is not noise but voice. In step S43, the recognition result for the second candidate recognition word is displayed, and the process ends.

In step S44, step S41
Then, when the determination value is smaller than Th2, it is determined that the input at the time of voice recognition is noise. In step S45, the recognition result when the recognition result is a noise word is displayed, and the process ends.

FIG. 5 is a diagram for explaining a concrete example. As shown in the configuration of the recognition dictionary in the voice recognition unit 5 of FIG. 9A, the recognition words of the recognition dictionary are 10 as the recognition dictionary of the voice recognition.
Five noise words (n) are registered in the word (m),
It is assumed that up to the third candidate (k) is output as the recognition result.
Furthermore, as shown in recognition result example 1 of FIG. 6D, when noise is input during voice recognition, the first candidate is the recognition word Suzuki, and the second and third candidates are noise words. explain.

When the first candidate is a recognition word, the second candidate and the third candidate
It is determined whether or not the candidate is a noise word, and the determination value is calculated. In the above-mentioned example 1, since the second candidate and the third candidate are both noise words, the judgment values w1 [2] (= 2) and the values w1 [1] (= shown in FIG. The judgment value becomes 3 by adding 1).

Since the first candidate of the recognition result is Suzuki, which is the recognition word, because the value is equal to or greater than the judgment value Th1 (= 3), the input at the time of voice recognition is judged to be noise, and the recognition result is recognized for the noise word. The result is displayed to prevent misrecognition. Next, as shown in (e) recognition result example 2 of this figure, an example in which when Suzuki is uttered during voice recognition, the first candidate is a noise word, the second candidate is Suzuki, and the third candidate is Sato To do.

When the first candidate is a noise word, it is determined whether the second candidate and the third candidate are recognition words, and the determination value is calculated. In the above-mentioned example 2, since the second candidate and the third candidate are both recognition words, the judgment weighting value w2 in FIG.
[2] (= 2) and w2 [1] (= 1) are added, and the determination value becomes 3.

Since the judgment value is a value of Th2 (= 3) or more, the first candidate of the recognition result is a noise word, but the input at the time of voice recognition is judged to be voice, and the second candidate Suzuki as the recognition result. Is output. In this way, since the determination process of the recognition word or the noise word is performed based on the appearance ratio of the noise word candidates and the appearance ratio of the recognition word candidates, erroneous recognition can be reduced.

Although the voice recognition system for the portable telephone device according to the embodiment of the present invention has been described above, the present invention is not limited to this embodiment, and various modifications can be made according to the spirit of the invention. . Therefore, according to the present invention, it is possible to prevent erroneous recognition and output a more accurate voice recognition result. As a result, the recognition performance of voice recognition can be improved.

Not only the first candidate of the recognition result but also the second candidate ~
This is because by using a plurality of recognition results up to the kth candidate, it is determined whether the input during voice recognition is voice or noise, and the recognition result is output. Next, another embodiment of the present invention will be described.
In the present embodiment, as a processing unit for determining whether the input during voice recognition is voice or noise, in addition to the processing for determining whether the second candidate to the kth candidate are noise or voice, the second candidate -When the recognition word is included up to the kth candidate, whether the number of characters of the recognition word is long or short is added as a determination value.

For example, when the recognition result candidate includes a recognition word, if the number of characters of the recognition word is small, the input at the time of voice recognition is likely to be noise, and if the number of characters is long, it is likely to be voice. It can be judged to be high. In this way, when the recognition result candidate includes a recognition word in addition to the determination as to whether the recognition result is speech or noise, the number of characters is also added to the determination value to output a more accurate speech recognition result. It becomes possible to do.

[0044]

As described above, according to the present invention,
By using not only the first candidate of the recognition result but also the plurality of recognition results of the second candidate to the k-th candidate, it is determined whether the input during voice recognition is voice or noise, and the recognition result is output. By doing so, erroneous recognition can be prevented and accurate voice recognition is possible.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a voice recognition system for a mobile phone device according to the present invention.

FIG. 2 is a flowchart illustrating an outline of operations of a main CPU 4, a voice recognition unit 5, and a recognition determination unit 4A in FIG.

FIG. 3 is a flowchart illustrating a detailed process of determining a recognition result in a recognition determining unit 4A of the main CPU 4 of FIG.

FIG. 4 is a flowchart illustrating a detailed process of determining a recognition result in a recognition determination unit 4A of the main CPU 4 of FIG.

FIG. 5 is a diagram illustrating a specific example.

[Explanation of symbols]

1 ... Wireless section 2 ... Operation part 3 ... Display 4 ... Main CPU 4A ... Recognition determination unit 5 ... Voice recognition unit 6 ... A / D converter 7 ... D / A converter 8 ... Microphone 9 ... Speaker 10 ... Voice synthesizer

Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G10L 15/20

Claims (4)

(57) [Claims] 1. A voice recognition system for a mobile phone device for voice input, comprising: holding a dictionary in which a plurality of recognition words indicating voice and a plurality of noise words are registered, and recognizing the voice input to select a plurality of candidates. A speech recognition unit that outputs a recognition result, and an appearance ratio of recognition word candidates recognized by the speech recognition unit,
A process of determining whether the word is a recognition word or a noise word is performed based on the appearance ratio of noise word candidates, and the first candidate is a recognition word and the second and subsequent candidates are recognized.
If the noise word occurrence ratio of is greater than or equal to the judgment value, the recognition result
The result is determined to be a noise word, and the noise words of the second and subsequent candidates are determined.
If the appearance ratio is less than the judgment value, the recognition result is regarded as the recognition word.
A voice recognition system for a mobile phone device, comprising a recognition determination unit for determination . 2. Voice recognition of a mobile phone device for voice input
In the system, multiple recognition words indicating voice and multiple noise words are registered.
Hold a dictionary, recognize the voice input and recognize multiple candidates.
A voice recognition unit that outputs a recognition result, and an appearance ratio of recognition word candidates recognized by the voice recognition unit,
Occurrence rate of noise word candidates, whether it is a recognition word or noise word
The first candidate is a noise word, and the second candidate
If the appearance ratio of the descending recognition word is more than the judgment value, the recognition result
The result is judged as a recognition word, and the recognition words after the second candidate appear.
If the ratio is less than the judgment value, the recognition result is judged as a noise word.
A voice recognition system for a mobile phone device, comprising: 3. The appearance ratio of the noise word and the recognition word
Occurrence rate is different for multiple noise words and multiple recognition words.
Each time a noise word or a recognition word appears as a candidate with a weight
Is calculated by adding each weight to
The voice recognition system for a mobile phone device according to claim 1 or 2 . 4. A voice recognition of a mobile phone device for voice input.
In the system, multiple recognition words indicating voice and multiple noise words are registered.
Hold a dictionary, recognize the voice input and recognize multiple candidates.
A voice recognition unit that outputs a recognition result, and an appearance ratio of recognition word candidates recognized by the voice recognition unit,
Occurrence rate of noise word candidates, whether it is a recognition word or noise word
Of the recognition word candidates,
The recognition word candidates composed of a smaller number of characters are
And a recognition determination unit that is a candidate for a noise word.
A voice recognition system for mobile phone devices.