JP6996944B2 - Speech recognition system - Google Patents

Description

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

As a voice recognition technique for recognizing a user's uttered voice, a technique for constantly recognizing a user's uttered voice without requiring a user's voice recognition start instruction operation is known (for example, Patent Document 1). ..

In addition, as a voice recognition technology for recognizing a user's voice, the terminal recognizes the user's voice by a voice recognition device provided by the terminal and also recognizes the user's voice by using an external voice recognition server. There is known a technique for performing voice recognition in the above and using the voice recognition result obtained by either voice recognition (for example, Patent Documents 2 and 3).

International Publication No. 2014/068788 JP-A-2015-102795 JP 2013-064777

When using the voice recognition of the user's spoken voice by the voice recognition device provided by the terminal and the voice recognition of the user's spoken voice by using an external voice recognition server for command input of different functions, they are performed in parallel. It is not appropriate to perform both voice recognition because the command input by the user by the spoken voice may be input as a command for a function different from the function intended by the user to input the command.

Therefore, it is desirable to selectively switch between the voice recognition of the user's spoken voice by the voice recognition device provided by the user and the voice recognition of the user's spoken voice by using the external voice recognition server.

On the other hand, when the voice recognition by the voice recognition device provided by the user is always performed without requiring the user's voice recognition start instruction operation, the voice recognition of the user's spoken voice by the voice recognition device provided by the user is performed. And, if the voice recognition of the user's spoken voice using an external voice recognition server is selectively switched, the following problems occur.

That is, in this case, the first mode in which only the voice recognition by the voice recognition device provided by the user is always performed without requiring the user's voice recognition start instruction operation, and the user using only the external voice recognition server. A second mode is provided for voice recognition of the spoken voice of the above, and normally, voice recognition is performed in the first mode, and voice recognition is temporarily performed in the second mode according to the user's operation, and voice recognition is performed. When is finished, it is conceivable to return to the first mode, but if this is done, the function of using the voice recognition by the voice recognition device provided by itself for inputting the command during the period of the second mode. Even if it becomes necessary to input an urgent command for, the second mode is forcibly switched to the first mode because there is no voice recognition start instruction operation for voice recognition by the first mode. This makes it impossible to enter the urgent command.

On the other hand, if voice recognition is performed by the voice recognition device provided by the user even during the period of the second mode, as described above, the command input by the user by the spoken voice is input by the user. It may be entered as a command for a function that is different from the intended function.

Therefore, the present invention is a voice recognition system in which a first voice recognition means and a second voice recognition means, which constantly perform voice recognition without requiring a user's voice recognition start instruction operation, are used for input to different functions. In the present invention, it is an object of the present invention to be able to perform urgent input using the voice recognition of the first voice recognition means at all times while suppressing erroneous input for each function.

In order to achieve the above object, the present invention is a first voice recognition in which a voice recognition system that recognizes a voice spoken by a user is constantly voice-recognized with a microphone and a command represented by the voice picked up by the microphone. The means, a first functional unit that executes a process in which an input command commands execution, a second voice recognition means that performs a voice recognition operation that recognizes the voice picked up by the microphone, and the second voice recognition means. A second functional unit that processes the voice recognition result of the voice recognition device, a priority command storage means that registers some commands among the commands that the first voice recognition means recognizes as voice recognition means, and voice recognition. It is provided as an operation control means. Here, the voice recognition operation control means controls switching between the first voice recognition mode and the second voice recognition mode provided as the voice recognition mode, and the voice recognition mode is the first voice. In the recognition mode, the voice recognition operation of the second voice recognition means is stopped, and the command recognized by the first voice recognition means is input to the first function unit to perform the voice recognition mode. Is the second voice recognition mode, the second voice recognition means is made to execute the voice recognition operation, and the command recognized by the first voice recognition means is used as the priority command storage means. Only when it is a registered priority command, the command recognized by the first voice recognition means is input to the first function unit.

Here, such a voice recognition system sets the voice recognition mode to the first voice recognition mode when a predetermined input from the user is generated while the voice recognition operation control means is in the first voice recognition mode. When the voice recognition operation of the second voice recognition means is completed while switching to the voice recognition mode of 2 and in the second voice recognition mode, the voice recognition mode is changed to the first voice recognition mode. It may be configured to switch to.

Further, in such a voice recognition system, the second voice recognition means is connected to an external voice recognition server that provides a voice recognition service as the voice recognition operation via communication, and the connected voice recognition server. The voice recognition service may be used to perform voice recognition of the voice picked up by the microphone.

Further, in such a voice recognition system, the voice recognition system may be a voice recognition system used for voice input in a system mounted on an automobile.
Further, such a voice recognition system is composed of an in-vehicle system mounted on an automobile and a portable device selectively connected to the in-vehicle system, and the in-vehicle system includes the microphone and the first voice recognition. The means, the first functional unit, the priority command storage means, and the voice recognition operation control means are provided, and the portable device is provided with the second voice recognition means and the second functional unit, and the voice recognition is performed. In the operation control means, when the voice recognition mode is in the second voice recognition mode, the voice picked up by the microphone is transferred from the in-vehicle system to the portable device, and the second voice recognition means. It may be configured to perform a voice recognition operation for recognizing the voice transferred from the in-vehicle system to the portable device.

Further, in each of the above voice recognition systems, among the commands for voice recognition of the priority command storage means by the first voice recognition means, the processing in which the command commands execution is related to ensuring the safety of the automobile. It may be assumed that the command is at least registered as a preferred command.

Further, in each of the above voice recognition systems, the priority command storage means is subjected to significant processing for the voice recognition result of voice recognition of the voice representing the command among the commands for voice recognition by the first voice recognition means. A command not specified in the second functional unit at that time may be registered as at least a priority command.

According to the voice recognition system as described above, in the first voice recognition mode, when the command is input to the first functional unit by the first voice recognition means, the voice recognition by the second voice recognition means is performed. When the voice is stopped and the voice is input to the second functional unit by the second voice recognition means in the second voice recognition mode, basically, the first voice recognition that constantly performs voice recognition is performed. Stops inputting commands to the first functional unit using the voice recognition of the means.

Therefore, the voice spoken by the user to input a command to the first function unit may be voice-input to the second function unit, or the voice spoken by the user to input voice to the second function unit. Is erroneously recognized as the voice of the command input to the first functional unit, and the erroneous input of the command to the first functional unit is suppressed.

On the other hand, even when the voice is input to the second functional unit by the second voice recognition means, the priority command registered in the priority command storage means is recognized by the voice recognition of the first voice recognition means. If so, this is input to the first functional unit.

Therefore, by registering a command for instructing the execution of an urgent process such as a command for instructing the execution of the process related to ensuring the safety of the automobile as a priority command in the priority command storage means, the first command is always performed. It becomes possible to input a command for instructing the execution of an urgent process using the voice recognition of the voice recognition means to the first functional unit.

Further, the command of the first functional unit, that is, the command of the second functional unit, whose significant processing for the voice recognition result of voice-recognizing the voice representing the command is not specified in the second functional unit at that time. By registering commands that do not hinder the voice input of the above as priority commands in the priority command storage means, the voice recognition of the first voice recognition means is always used for these commands. You will be able to input commands to the first functional part.

As described above, according to the present invention, the first voice recognition means and the second voice recognition means, which constantly perform voice recognition without requiring the user's voice recognition start instruction operation, are input to different functions. In the voice recognition system used for the above, it is possible to always perform urgent input using the voice recognition of the first voice recognition means while suppressing erroneous input for each function.

It is a block diagram which shows the structure of the information processing system which concerns on embodiment of this invention. It is a flowchart which shows the external voice recognition control processing which concerns on embodiment of this invention. It is a flowchart which shows the speech recognition result filter processing which concerns on embodiment of this invention. It is a sequence diagram which shows the operation example of the information processing system which concerns on embodiment of this invention.

Hereinafter, embodiments of the present invention will be described by taking as an example an application to an information processing system used in an automobile.
FIG. 1 shows the configuration of the information processing system.
As shown in the figure, the information processing system includes an in-vehicle system 1 mounted on an automobile and a portable device 2 selectively connected to the in-vehicle system 1.
Here, the portable device 2 is a device that can be carried by a user such as a smartphone or a tablet device. Further, the portable device 2 is connected to an external voice recognition server 3 via mobile communication, and uses the voice recognition service of the voice recognition server 3 to perform voice recognition of the voice transferred from the in-vehicle system 1 to perform voice recognition. It has a function of accepting the result of the above as a voice input to the portable device 2 and performing an operation according to the voice input.

Next, the in-vehicle system 1 communicates with the microphone 101, the voice recognition device 102, the voice recognition dictionary 103, the voice recognition result filter unit 104, the priority word table 105, the external voice recognition control unit 106, the talk switch 107, and the portable device 2. It is provided with various peripheral devices 110 such as a communication interface 108, a data processing device 109, a display, a camera for photographing the periphery of an automobile, an AV device, and an air conditioning device.

Then, in the voice recognition dictionary 103, data for voice recognition of a word representing a command of the data processing device 109 is registered. Then, when the voice recognition device 102 uses the voice recognition dictionary 103 and the voice spoken by the user input from the microphone 101 is a voice representing a command of the data processing device 109, the voice recognition device 102 recognizes the command as a voice recognition result. The operation of outputting to the voice recognition result filter unit 104 is always performed without being triggered by the user's voice recognition start instruction operation.

Next, in the priority word table 105, among the commands represented by the words whose voice recognition data is registered in the voice recognition dictionary 103, the commands that need to be processed urgently and the voice input to the portable device 2 are performed. A command that never appears is registered. Then, the voice recognition result filter unit 104 performs the voice recognition result filter processing described in detail later using the priority word table 105, and satisfies a predetermined condition among the voice recognition results recognized by the voice recognition device 102. Only the voice recognition result is output to the data processing device 109.

The command that needs to be processed urgently is, for example, a command that instructs the data processing device 109 to execute a process related to ensuring the safety of the vehicle. For example, the data processing device 109 takes a picture of the rear of the vehicle. The command "back camera" or the like that instructs the display of the image taken by the camera on the display can be a command that needs to be processed urgently.

Further, as a command that does not appear in the voice input to the portable device 2, for example, when the portable device 2 uses the voice input to input the command to the portable device 2 at that time, the voice recognition dictionary 103 has voice. Among the commands represented by the words in which the recognition data is registered, the commands that the portable device 2 does not support can be used.

Further, among the commands represented by the words in which the data for voice recognition is registered in the voice recognition dictionary 103, the commands that are unlikely to appear in the voice input to the portable device 2 are also the voice input to the portable device 2. It may be a command that does not appear in.

Here, as a command that does not appear in the voice input to such a portable device 2, for example, there is a command "air volume up" that instructs the data processing device 109 to execute the air volume increasing process of the air conditioner.

Then, the data processing device 109 performs processing according to the command output from the voice recognition result filter unit 104.
Further, the external voice recognition control unit 106 performs the external voice recognition control process described in detail later to transfer the user's spoken voice input from the microphone 101 to the portable device 2 via the communication interface 108, and the portable device 2 To perform voice recognition using the voice recognition service of the voice recognition server 3 described above.

Hereinafter, the external voice recognition control process performed by the external voice recognition control unit 106 will be described.
FIG. 2 shows the procedure of the external voice recognition control process.
As shown in the figure, in the external voice recognition control process, the external voice recognition control unit 106 monitors that the talk switch 107 is turned on by the user (step 202), and if it is turned on, the external voice recognition mode is set. Set (step 204).

Then, a voice recognition start command is issued to the portable device 2 via the communication interface 108 (step 206), and transfer of the user's spoken voice input from the microphone 101 to the portable device 2 via the communication interface 108 is started (step 206). Step 208).

Here, when the portable device 2 receives the voice recognition start command from the vehicle-mounted system 1, the portable device 2 is a series of voices whose end point is the start point of the silent section having a predetermined time length or longer, which is transferred from the vehicle-mounted system 1. The voice recognition process for performing voice recognition using the voice recognition service of the voice recognition server 3 is started.

Next, if the external voice recognition control unit 106 starts transferring the user's spoken voice to the portable device 2 via the communication interface 108 (step 208), the external voice recognition stop from the voice recognition result filter unit 104. The occurrence of the reception of the command (step 210) and the reception of the voice recognition end notification from the portable device 2 via the communication interface 108 (step 212) are monitored.

Here, the portable device 2 outputs a voice recognition end notification to the in-vehicle device when the above-mentioned voice recognition process is completed.
Then, the external voice recognition control unit 106 receives either a voice recognition stop command from the voice recognition result filter unit 104 (step 210) or a voice recognition end notification from the portable device 2 (step 212). When the above occurs, the transfer of the user's spoken voice input from the microphone 101 to the portable device 2 via the communication interface 108 is stopped (step 214), and the external voice recognition mode is canceled (step 216).

Then, the process returns to the process from step 202.
The external voice recognition control process performed by the external voice recognition control unit 106 has been described above.
Next, the above-mentioned voice recognition result filter processing performed by the voice recognition result filter unit 104 will be described.
FIG. 3 shows a procedure for voice recognition result filtering.
As shown in the figure, the voice recognition result filter unit 104 waits for the input of the voice recognition result from the voice recognition device 102 (step 302) in the voice recognition result filter processing, and if the voice recognition result is input, the external voice recognition control. Check whether the external voice recognition mode is set by the unit 106 (step 304).

If the external voice recognition mode is not set (step 304), the input voice recognition result is output to the data processing device 109 (step 310), and the process returns to the process from step 302.

On the other hand, when the external voice recognition mode is set (step 304), it is checked whether the input voice recognition result is a command registered in the priority word table 105 (step 306).

If the input voice recognition result is not a command registered in the priority word table 105 (step 306), the received voice recognition result is discarded and the process returns to the process from step 302 as it is.

On the other hand, if the received voice recognition result is a command registered in the priority word table 105 (step 306), the external voice recognition stop command is transmitted to the external voice recognition control unit 106 (step 308). The received voice recognition result is output to the data processing device 109 (step 310), and the process returns to the process from step 302.

The voice recognition result filter processing performed by the voice recognition result filter unit 104 has been described above.
Here, FIG. 4 shows an example of the operation of voice recognition by the above-mentioned external voice recognition control processing and voice recognition result filter processing.
As shown in the figure, normally, the voice spoken by the user input from the microphone 101 is sent to the voice recognition device 102 (401), voice recognition is performed by the voice recognition device 102, and the voice recognition result is sent to the voice recognition result filter unit 104. Sent (402). Then, the voice recognition result filter unit 104 outputs the received voice recognition result to the data processing device 109 (403).

On the other hand, when the user turns on the talk switch 107 in order to input voice to the portable device 2 (411), the external voice recognition control unit 106 sets the external voice recognition mode (412).

Then, when the user utters a word that is not a word representing the command registered in the priority word table 105, the voice (unregistered word voice) is transmitted from the microphone 101 to the voice recognition device 102 and the external voice recognition control unit 106. Is sent to (413).

The voice recognition device 102 recognizes the received voice and sends the voice recognition result to the voice recognition result filter unit 104 (414). Since the received voice recognition result is not a command registered in the priority word table 105, the voice recognition result filter unit 104 discards the voice recognition result without outputting it to the data processing device 109.

On the other hand, the external voice recognition control unit 106 transfers the received voice to the portable device 2 (415), and the portable device 2 transmits the transferred voice to the voice recognition server 3 (416).
Then, after that, before the portable device 2 receives the voice recognition result of the transmitted (416) voice from the voice recognition server 3, when the user speaks a word representing the command registered in the priority word table 105, the voice ( The registered word voice) is sent from the microphone 101 to the voice recognition device 102 and the external voice recognition control unit 106 (417).

The voice recognition device 102 recognizes the received voice and sends the voice recognition result to the voice recognition result filter unit 104 (418). Since the received voice recognition result is a command registered in the priority word table 105, the voice recognition result filter unit 104 issues an external voice recognition stop command to the external voice recognition control unit 106 (419), and the voice recognition result. Is output to the data processing device 109 (420).

On the other hand, the external voice recognition control unit 106 transfers the received voice to the portable device 2 (421). Here, it is assumed that the portable device 2 is configured to ignore the transferred voice during the period of waiting for the voice recognition result from the voice recognition server 3, and is transferred to the portable device 2 (). 421) The voice shall be discarded without being transmitted to the voice recognition server 3.

Further, when the external voice recognition control unit 106 receives the external voice recognition stop command (419), the external voice recognition control unit 106 cancels the external voice recognition mode (422).
On the other hand, when the voice recognition result of the (416) voice transmitted by the portable device 2 is responded to the portable device 2 from the voice recognition server 3 (423), the portable device 2 processes the voice recognition result.

Further, the voice recognition process of the portable device 2 is completed, and the portable device 2 notifies the external voice recognition control unit 106 of the end of voice recognition (424).
The above is an example of the operation of voice recognition by the external voice recognition control processing and the voice recognition result filter processing.
Here, unlike the example shown in FIG. 4, immediately after the user turns on the talk switch 107, or after the user turns on the talk switch 107, a word that is not a word representing a command registered in the priority word table 105 is inserted in the middle. Even when a word representing a command registered in the priority word table 105 is spoken after the utterances up to, the recognition result of recognizing the utterance of the word by the voice recognition device 102 is processed by data processing via the voice recognition result filter unit 104. It will be output to the device 109.
In these cases, the spoken voice of the word representing the command registered in the priority word table 105 is transferred from the external voice recognition control unit 106 to the portable device 2, and the portable device 2 uses the voice recognition server 3 for the voice. Inconvenience that voice recognition is performed may occur, but if the voice spoken by the user represents a command that needs to be processed urgently, the input of the command to the data processing device 109 is prioritized over the inconvenience. As long as the voice spoken by the user represents a command that does not appear in the voice input to the portable device 2, the portable device 2 does not perform an unexpected operation based on the voice recognition result.

The embodiment of the present invention has been described above.
According to the present embodiment, normally, the voice recognition device 102 constantly recognizes the voice of the user and voices by the voice recognition device 102 in a state where the voice input of the voice of the user using the voice recognition server 3 by the portable device 2 is stopped. The recognition result command is input to the data processing device 109, but when the user turns on the talk switch 107 to input voice to the portable device 2, the voice recognition server 3 by the portable device 2 is used. The voice input of the spoken voice of the existing user is started.

Then, during the period during which the voice input of the user's spoken voice is performed using the voice recognition server 3 by the portable device 2, basically, the command of the voice recognition result of the user's spoken voice by the voice recognition device 102 is performed. The input to the data processing device 109 is stopped.

Therefore, the voice spoken by the user to input a command to the data processing device 109 is input to the portable device 2, or the voice spoken by the user to input the voice to the portable device 2 is input to the data processing device. It is suppressed that the command is erroneously input to the data processing device 109 by erroneously recognizing it as the voice of the command input to the 109.

On the other hand, even during the period during which the voice input of the user's spoken voice using the voice recognition server 3 by the portable device 2 is being performed, it is requested to execute the urgent processing registered in the priority word table 105. Only when the user utters a command to be executed or a word representing a command that does not appear in the voice input to the portable device 2, the command recognized by the voice recognition device 102 is input to the data processing device 109.

Therefore, according to the present embodiment, while suppressing erroneous input of a command to the data processing device 109 and erroneous voice input to the portable device 2, the voice of a command that constantly requests the data processing device 109 to execute an urgent process. You will be able to input. Further, even for a command that does not appear in the voice input to the portable device 2, the voice input to the data processing device 109 can be performed at all times.

In the above embodiment, the on operation of the talk switch 107 is a trigger for causing the portable device 2 to perform voice recognition, but this trigger may be something other than the on operation of the talk switch 107. That is, this trigger may be a voice input of a command instructing the start of a voice input to the portable device 2. In this case, the generation of the voice input of the command instructing the start of the voice input to the portable device 2 is detected by the voice recognition device 102 by voice recognition of the word representing the command in the voice of the user. ..

Further, in the above embodiment, the voice using the voice recognition server 3 for the spoken voice input from the microphone 101 is used instead of the portable device 2 for performing voice recognition using the voice recognition server 3 for the spoken voice input from the microphone 101. Even if it is equipped with a device that is not a portable device 2 that performs recognition, or an arbitrary device that performs voice recognition of spoken voice input from the microphone 101 using its own voice recognition function without using the voice recognition server 3. , The portable device 2 can be similarly applied by replacing the device with the device.

1 ... In-vehicle system, 2 ... Portable device, 3 ... Voice recognition server, 101 ... Microphone, 102 ... Voice recognition device, 103 ... Voice recognition dictionary, 104 ... Voice recognition result filter unit, 105 ... Priority word table, 106 ... External voice Recognition control unit, 107 ... talk switch, 108 ... communication interface, 109 ... data processing device, 110 ... peripheral device.

Claims (7)

It is a voice recognition system that recognizes the voice spoken by the user.
With a microphone
A first voice recognition means for constantly recognizing a command represented by a voice picked up by the microphone, and a first voice recognition means.
The first functional part that executes the process in which the input command commands execution, and
A second voice recognition means that performs a voice recognition operation that recognizes the voice picked up by the microphone, and
A second functional unit that processes the voice recognition result of the second voice recognition means , and
The priority command storage means in which some of the commands recognized by the first voice recognition means are registered as priority commands, and the priority command storage means.
It has a voice recognition operation control means and
The voice recognition operation control means is
While controlling the switching between the first voice recognition mode and the second voice recognition mode provided as the voice recognition mode,
When the voice recognition mode is the first voice recognition mode, the voice recognition operation of the second voice recognition means is stopped, and the command that the first voice recognition means recognizes by voice is the first. Enter in the function part,
When the voice recognition mode is the second voice recognition mode, the command that causes the second voice recognition means to execute the voice recognition operation and the first voice recognition means recognizes the voice is given priority. A voice recognition system characterized in that a command recognized by the first voice recognition means is input to the first functional unit only when the priority command is registered in the command storage means. The voice recognition system according to claim 1.
The voice recognition operation control means switches the voice recognition mode to the second voice recognition mode when a predetermined input from the user occurs while in the first voice recognition mode, and the second voice recognition mode. A voice recognition system comprising switching the voice recognition mode to the first voice recognition mode when the voice recognition operation of the second voice recognition means is completed while in the voice recognition mode. The voice recognition system according to claim 1 or 2.
As the voice recognition operation, the second voice recognition means connects to an external voice recognition server that provides a voice recognition service via communication, and uses the voice recognition service of the connected voice recognition server to use the microphone. A voice recognition system characterized by performing voice recognition of the voice picked up by. The voice recognition system according to claim 1, 2 or 3.
The voice recognition system is a voice recognition system characterized by being a voice recognition system used for voice input in a system mounted on an automobile. The voice recognition system according to claim 1, 2 or 3.
It has an in-vehicle system mounted on an automobile and a portable device selectively connected to the in-vehicle system.
The in-vehicle system includes the microphone, the first voice recognition means, the first functional unit, the priority command storage means, and the voice recognition operation control means, and the portable device includes the second voice recognition. It has means and the second functional unit, and has
The voice recognition operation control means transfers the voice picked up by the microphone from the in-vehicle system to the portable device when the voice recognition mode is in the second voice recognition mode.
The second voice recognition means is a voice recognition system characterized by performing a voice recognition operation for voice recognition of voice transferred from the vehicle-mounted system to the portable device. The voice recognition system according to claim 4 or 5 .
The priority command storage means includes at least a priority command among the commands for voice recognition by the first voice recognition means, in which the process in which the command commands execution is a process related to ensuring the safety of the vehicle. A speech recognition system characterized by being registered as. The voice recognition system according to claim 1, 2, 3, 4, 5 or 6.
In the priority command storage means, among the commands for which the first voice recognition means recognizes the voice, significant processing for the voice recognition result of voice recognition of the voice representing the command is performed in the second functional unit at that time. A speech recognition system characterized in that commands not specified in are registered as at least priority commands.

Images