JP2020129130A - Information processing device - Google Patents

Abstract

To provide an information processing device by which an appropriate recognition result can be obtained.SOLUTION: A CPU 13 in a car navigation system 10 obtains a recognition result of a voice recognition engine 12 that performs voice recognition of voice made by a user and parameters made up of sound pressure information, a score, and algorithm determination. The CPU 13 further obtains a recognition result of a voice recognition engine 22 in a smart phone 20 that has performed voice recognition of the same voice, and parameters made up of sound pressure information, a score, and algorithm determination via a short distance wireless communication unit 15. The CPU 13 selects either one of the recognition result of the voice recognition engine 12 and the recognition result of the voice recognition engine 22 on the basis of the parameters of the voice recognition engine 12 and the parameters of the voice recognition engine 22, and causes the one to be executed as a command.SELECTED DRAWING: Figure 2

Description

The present invention relates to an information processing device that performs information processing based on a voice recognition processing result.

2. Description of the Related Art Conventionally, there are various electronic devices that recognize a voice uttered by a user and perform an operation according to the recognition result. Since the algorithms and dictionaries of the voice recognition processing unit are different for each device, the recognition rate is different for each device even for the same voice.

In recent years, cloud-type voice recognition systems have been proposed. In this system, a server is equipped with a voice recognition engine and a dictionary, and a plurality of terminals respectively connect to the server via a network to request voice recognition processing and acquire the processing result. Since such a cloud-type speech recognition system has a dictionary in the server, it has the advantages that the vocabulary can be greatly increased and that even advanced algorithms can be processed.

Further, Patent Document 1 discloses sharing the result of voice recognition with another electronic device. Specifically, when inputting to the external device 20 by voice, the recognition module is transmitted from the portable information processing device 10 to the external device 20, and the external device 20 performs processing using the received recognition module.

JP, 2003-140690, A

The cloud-type voice recognition system basically recognizes a conversation. However, for example, in a car navigation system, a unique phrase such as detour search or reroute can be used as a command for voice operation, but the cloud-type voice recognition system can use the phrase used in such a unique environment. Proper recognition is difficult.

Further, in the method described in Patent Document 1, although the recognition result can be shared by a plurality of electronic devices, for example, when the recognition module of the car navigation system is moved to another device, the recognition module is unique to the car navigation system. In order to perform recognition suitable for the environment, for example, when recognition of a normal conversation is performed, an appropriate recognition result may not be obtained.

Therefore, in view of the above-described problems, it is an object of the present invention to provide, for example, an information processing device that can obtain an appropriate recognition result.

In order to solve the above problems, a first acquisition unit that acquires first voice recognition result information and first voice recognition processing information from a first voice recognition unit that recognizes a voice, and a second voice recognition unit that recognizes the voice. A second acquisition unit that acquires second voice recognition result information and second voice recognition processing information from the first voice recognition result information or the second voice recognition processing information based on the first voice recognition processing information and the second voice recognition processing information. A control unit that selects any one of the two voice recognition result information and causes a processing unit to execute a process related to the selected first voice recognition result information or the second voice recognition result information. ..

According to a twelfth aspect of the present invention, there is provided a control method of an information processing device for causing a processing unit to perform processing based on a recognized voice, wherein the first voice recognition unit for recognizing the voice outputs first voice recognition result information and A first acquisition step of acquiring first voice recognition processing information; a second acquisition step of acquiring second voice recognition result information and second voice recognition processing information from a second voice recognition unit that recognizes the voice; One of the first voice recognition result information and the second voice recognition result information is selected based on the first voice recognition process information and the second voice recognition process information, and the selected first voice recognition result information or the first voice recognition result information. And a control step of causing the processing unit to execute processing relating to two voice recognition result information.

The invention described in claim 13 is characterized in that a computer executes the information processing method according to claim 12.

The invention according to claim 14 is characterized in that the information processing program according to claim 13 is stored.

1 is an external perspective view of a car navigation system according to an embodiment of the present invention. FIG. 2 is a block configuration diagram of a car navigation system and a smartphone shown in FIG. 1. 3 is a flowchart of a voice recognition operation of the car navigation system shown in FIG. 4 is a table illustrating a specific example of the flowchart shown in FIG. 3. 4 is a table illustrating a specific example of the flowchart shown in FIG. 3.

Hereinafter, an information processing apparatus according to an embodiment of the present invention will be described. In the information processing apparatus according to the embodiment of the present invention, the first acquisition unit recognizes the voice uttered by the user as voice recognition result information, which is a voice recognition result of the first voice recognition unit; A second voice that recognizes the voice recognized by the first voice recognition unit, and the first voice recognition processing information that is information obtained from the voice recognition unit together with the first voice recognition result information. The second voice recognition result information which is the voice recognition result of the recognition unit and the second voice recognition processing information which is the information obtained from the second voice recognition unit together with the second voice recognition result information are acquired. Then, the control unit, based on the first voice recognition process information acquired by the first acquisition unit and the second voice recognition process information acquired by the second acquisition unit, the first voice recognition result information or the second voice recognition result information. One of the two is selected, and the processing unit is caused to execute information processing based on the selected first voice recognition result information or second voice recognition result information. By doing so, it is possible to select from the results of the two voice recognition units that have recognized the same voice, so that it is possible to perform voice recognition with accuracy higher than that of performing voice recognition independently. For example, if the two voice recognition units have different algorithms and dictionaries, recognition results suitable for various environments can be obtained. Therefore, an appropriate recognition result can be obtained.

Further, when the first voice recognition processing information is equal to or more than a first threshold value set in advance, the control unit selects the first voice recognition result information and processes information processing based on the first voice recognition result information. It may be executed by a department. By doing so, various information processing such as the route search of the navigation system and the search of the store using the Internet can be performed by using the recognition result of the first voice recognition unit.

Further, when the first voice recognition processing information is less than the first threshold value and the second voice recognition processing information is more than a predetermined second threshold value, the control unit selects the second voice recognition result information, The processing unit may be caused to execute information processing based on the second voice recognition result information. By doing so, when the reliability of the recognition result of the first voice recognition unit is low and there is a high possibility that it is not appropriate, the result of the second voice recognition unit is used, for example, the route search of the navigation system or the Internet is used. It is possible to perform various types of information processing such as searching for a store or the like.

The second acquisition unit further acquires processing result information that is a result of processing based on the second voice recognition result information. Then, when the first voice recognition processing information is less than a predetermined first threshold value and the second voice recognition processing information is more than a predetermined second threshold value, the control unit determines that the second voice recognition result information has a predetermined value. It is determined whether or not it is included in the command group. If it is included, the processing unit is caused to execute information processing based on the second voice recognition result information. If it is not included, the second acquisition unit acquires it. The processing unit may be caused to execute information processing based on the processed result information. By doing so, when the recognition result of the first voice recognition unit is highly likely to be inappropriate, the result of the second voice recognition unit can be used. Further, when the result of the second voice recognition unit is included in a predetermined command group such as a command for operating the device, the operation according to the command can be performed. When the result is not included in the predetermined command group, information processing can be performed using the result processed by the device having the second voice recognition unit.

In addition, when the first voice recognition processing information is less than the first threshold value and the second voice recognition processing information is less than the second threshold value, the control unit provides the first voice recognition processing information and the second voice recognition processing information respectively. A predetermined weighted calculation may be performed, and the first voice recognition result information or the second voice recognition result information may be selected based on the calculation result. By doing so, when the recognition result cannot be selected with the first threshold value and the second threshold value, the first voice recognition result is calculated by performing a weighting operation on each result based on the usage environment. Either the information or the second voice recognition result information can be selected.

Further, when the first voice recognition processing information is less than the first threshold value and the second voice recognition processing information is less than the second threshold value, the control unit determines the first voice recognition result information or the first voice recognition result information based on the past usage history. Two voice recognition result information may be selected. By doing so, when the recognition result cannot be selected with the first threshold value and the second threshold value, the first voice recognition result information is obtained based on the past use history such as whether it has been used for past voice recognition, search or operation. Alternatively, either one of the second voice recognition result information can be selected.

Further, when the first voice recognition processing information is less than the first threshold value and the second voice recognition processing information is less than the second threshold value, the control unit determines the first voice recognition result information or the first voice recognition result information based on the past usage status. Two voice recognition result information may be selected. By doing so, when the recognition result cannot be selected by the first threshold value and the second threshold value, the first voice recognition result information or the second voice recognition result information based on the past usage situation such as time zone, season, and weather. Either one can be selected.

Further, the first voice recognition processing information and the second voice recognition processing information are information regarding the degree of association between the voice uttered by the user and the first voice recognition result information or the second voice recognition result information. It may include information. By doing so, the first voice recognition result information or the second voice recognition result information is based on the degree of association with the recognition result based on the uttered voice, that is, the score indicating the similarity between the uttered voice and the dictionary. Either one can be selected.

Further, the first voice recognition unit, the first acquisition unit, the second acquisition unit, and the control unit may be integrally provided. By doing so, an appropriate recognition result can be obtained in the information processing apparatus by linking with an external device or the like having the second voice recognition unit.

Further, the control unit may cause the first voice recognition unit and the second voice recognition unit to learn the recognition result based on the first voice recognition result information or the second voice recognition result information selected by itself. By doing so, the recognition results can be shared by the respective voice recognition units, and the accuracy of the subsequent voice recognition can be improved.

Further, an input unit for inputting each voice uttered by the user may be provided corresponding to each of the first voice recognition unit and the second voice recognition unit. By doing so, for example, a microphone can be provided as an input unit corresponding to each voice recognition unit, and the sound pressure of the voice input from the microphone as the first voice recognition processing information and the second voice recognition processing information can be provided. And the volume can be obtained.

Also, in the control method of the information processing apparatus according to the embodiment of the present invention, the first voice recognition result which is the voice recognition result of the first voice recognition unit that voice-recognizes the voice uttered by the user in the first acquisition step. The information and the first voice recognition processing information, which is the information obtained from the first voice recognition unit together with the first voice recognition result information, are acquired, and the voice recognized by the first voice recognition unit is voiced in the second acquisition step. The second voice recognition result information that is the voice recognition result of the second voice recognition unit to be recognized, and the second voice recognition processing information that is the information obtained from the second voice recognition unit together with the second voice recognition result information are acquired. .. Then, in the control step, the first voice recognition result information or the second voice recognition result information based on the first voice recognition processing information acquired in the first acquisition step and the second voice recognition processing information acquired in the second acquisition step. One of the two is selected, and the processing unit is caused to execute information processing based on the selected first voice recognition result information or second voice recognition result information. By doing so, it is possible to select from the results of the two voice recognition units that have recognized the same voice, so that it is possible to perform voice recognition with accuracy higher than that of performing voice recognition independently. For example, if the two voice recognition units have different algorithms and dictionaries, recognition results suitable for various environments can be obtained. Therefore, an appropriate recognition result can be obtained.

Further, it may be a control program of an information processing apparatus that causes a computer to execute the above-described method of controlling the information processing apparatus. By doing so, an appropriate recognition result can be obtained using a computer.

Further, the control program of the information processing apparatus described above may be stored in a computer-readable recording medium. By doing so, the program can be distributed as a single unit in addition to being installed in the device, and the version can be easily upgraded.

A car navigation system 10 as an information processing apparatus according to an embodiment of the present invention will be described with reference to FIGS. 1 to 5. The car navigation system 10 is mounted on an instrument panel 100 of the vehicle as shown in FIG. Then, the car navigation system 10 is capable of communicating by connecting to a smartphone 20, which will be described later, placed on the passenger seat 101, for example, wirelessly or by wire.

As shown in FIG. 2, the car navigation system 10 shown in FIG. 1 includes a microphone 11, a voice recognition engine 12, a CPU 13, a storage device 14, a short-range wireless communication unit 15, a GPS 16, and a display unit. 17 is provided.

The microphone 11 as an input unit receives the voice uttered by the user, converts the voice into a voice signal which is an electric signal, and outputs the voice signal to the voice recognition engine 12. The microphone 11 does not have to be provided integrally with the car navigation system 10, but may be provided near the user, for example, in a steering column or the like in the case of a vehicle, and connected by a cable or wireless.

The voice recognition engine 12 as the first voice recognition unit performs voice recognition based on the voice signal input from the microphone 11, and recognizes a word or a phrase that is a combination of words, which is the recognition result information, and the voice signal and itself. A score indicating the degree of approximation to a vocabulary in a dictionary that the user has, an algorithm determination that indicates the degree of determination by an algorithm such as the degree of matching with the dictionary that the user has, the degree of matching with the candidates estimated from the contexts before and after, input from the microphone 11. The parameters such as the sound pressure information of the audio signal and the like are output to the CPU 13. There may be a plurality of result information as candidates, and in that case, a score, algorithm determination, and sound pressure information are output for each candidate. Volume information may be used instead of sound pressure information. Further, the parameter may be one or more of the above-mentioned three, but it is preferable that the score is included. That is, the phrase corresponds to the first voice recognition result information and the parameter corresponds to the first voice recognition processing information. The score, which is the degree of approximation between the voice signal and the dictionary, is the degree-of-relevance information that is information regarding the degree of association between the voice signal and the phrase retrieved from the dictionary.

The recognition algorithm and the score calculation method used in the voice recognition engine 12 are well known and not particularly limited. However, since the voice recognition engine 12 is provided in the car navigation system 10, It is preferable to adjust the recognition probability of phrases such as "reroute" and "detour search" used for voice input commands (operation commands) to be high. Further, the voice recognition engine 12 may not be included in the car navigation system 10, and for example, the cloud-type voice recognition system described in the related art may be used. That is, the voice signal input from the microphone 11 may be transmitted to a server or the like, the server may perform voice recognition processing, and the car navigation system 10 may receive the phrase and the parameter.

The CPU 13 as the control unit, the first acquisition unit, and the processing unit is configured as a microcomputer including a RAM, a ROM, and the like, and controls the entire car navigation system 10. Then, the CPU 13 executes various functions that the car navigation system 10 generally has, such as destination setting, route search, guidance, and map display. Further, the CPU 13 acquires the phrase and the parameter output by the voice recognition engine 12. Then, based on the parameter output by the voice recognition engine 12 and the parameter output by the voice recognition engine 22 acquired by the short-range wireless communication unit 15 from the smartphone 20 described below, the phrase output by the voice recognition engine 12 and the proximity from the smartphone 20. One of the phrases output by the voice recognition engine 22 acquired by the distance wireless communication unit 15 is selected, and the process based on the selected phrase is executed.

The storage device 14 is composed of a non-volatile readable/writable storage medium such as a hard disk or a semiconductor memory. The storage device 14 stores, for example, information such as maps used for guidance in the car navigation system 10.

The short-range wireless communication unit 15 as the second acquisition unit is connected to the smartphone 20 described later by short-range wireless communication such as Bleuthooth (registered trademark) or infrared communication, and performs data communication with each other. The short-range wireless communication unit 15 also acquires from the smartphone 20 a phrase and a parameter output by a voice recognition engine 22 described later. The short-distance wireless communication unit 15 is not limited to short-distance wireless communication, and may be another wireless communication such as a wireless LAN (Local Area Network) or wired communication such as a USB (Universal Serial Bus).

As is well known, the GPS 16 receives radio waves transmitted from a plurality of GPS (Global Positioning System) satellites, obtains current position information (current position information), and outputs it to the CPU 13. In the present embodiment, the GPS 16 is provided integrally with the car navigation system 10, but the GPS 16 may be configured as a separate body and detachable from the car navigation system 10.

The display unit 17 is composed of a display device such as a liquid crystal display or an EL (Electro Luminescence) display. Further, the display unit 17 may have a touch panel overlaid on the display surface. The display unit 17 displays a map, the position of the vehicle, guidance information such as destinations and routes, various operation menus, buttons for operating the touch panel, and the like.

In the car navigation system 10 having the above-described configuration, the smartphone 20 and the short-range wireless communication unit 15 illustrated in FIG. 2 perform data communication with each other. As described above, the car navigation system 10 performs the navigation function such as route search by itself with the map information. However, the car navigation system 10 causes the server to execute the navigation function with the map information by an external server. May receive and display the result.

The smartphone 20 includes a microphone 21, a voice recognition engine 22, a CPU 23, a storage device 24, a short-range wireless communication unit 25, and a line communication unit 26.

The microphone 21 as an input unit receives the voice uttered by the user, converts the voice into a voice signal which is an electric signal, and outputs the voice signal to the voice recognition engine 22.

The voice recognition engine 22 as the second voice recognition unit performs voice recognition based on the voice signal input from the microphone 21, and recognizes a word or a phrase based on a combination of words, which is the recognition result information, and the voice signal and itself. The CPU 23 outputs to the CPU 23 a score indicating the degree of approximation to the vocabulary in the dictionary that it has, an algorithm determination that indicates the degree of matching with the dictionary that it has, and sound pressure information of the audio signal input from the microphone 21. There may be a plurality of result information as candidates, and in that case, a score, algorithm determination, and sound pressure information are output for each candidate. That is, the phrase corresponds to the second voice recognition result information and the parameter corresponds to the second voice recognition processing information.

The recognition algorithm used in the voice recognition engine 22 and the score calculation method are well known and are not particularly limited, but the voice recognition engine 22 preferably has a recognition algorithm and a dictionary different from those of the voice recognition engine 12. .. In this case, the possibility that a phrase that cannot be correctly recognized by the voice recognition engine 12 can be recognized increases, and the voice recognition engine 12 can be supplemented.

The voice recognition engine 22 may not be included in the smartphone 20, and for example, the cloud-type voice recognition system described in the related art may be used. That is, the voice signal input from the microphone 21 may be transmitted to the server or the like, the server may perform the voice recognition process, and the smartphone 20 may receive the phrase and the parameter.

The CPU 23 is configured as a microcomputer including a RAM, a ROM and the like, and controls the entire smartphone 20. Then, the CPU 23 executes a function that the smartphone 20 generally has, for example, a function such as a telephone call, a mail, an internet connection, or the like, or an application. Further, the CPU 23 acquires the phrase and the parameter from the voice recognition engine 22 and transmits the phrase and the parameter to the car navigation system 10 via the short-range wireless communication unit 25. In addition, processing such as Internet search is performed based on the recognition result of the voice recognition engine 22.

The storage device 24 is composed of a nonvolatile readable/writable storage medium such as a semiconductor memory. A removable storage medium such as a memory card may be used. The storage device 24 stores, for example, a telephone directory or application data used by the smartphone 20.

The short-range wireless communication unit 25 is connected to the car navigation system 10 described later by short-range wireless communication such as Bluetooth (registered trademark) or infrared communication to perform data communication with each other. The short-range wireless communication unit 25 also transmits the phrase and the parameter output by the voice recognition engine 22 to the car navigation system 10.

The line communication unit 26 connects to a mobile phone line network and performs various communications. The line communication unit 26 is connected to a base station or the like by a communication method such as W-CDMA (Wideband Code Division Multiple Access) or LTE (Long Term Evolution) to connect to a mobile phone line network.

The car navigation system 10 configured as described above selects a more appropriate recognition result (phrase) using both the voice recognition engine 12 of itself and the voice recognition engine 22 of the smartphone 20. Then, various processes based on the selected phrase are executed. The detailed operation will be described with reference to the flowchart shown in FIG. The flowchart shown in FIG. 3 is executed by the CPU 13. In addition, before executing this flowchart, the car navigation system 10 and the smartphone 20 can mutually perform data communication by short-range wireless communication.

First, in step S1 as the first acquisition step, the voice uttered by the user is voice-recognized by the voice recognition engine 12 (car navigation system 10) via the microphone 11, and the phrases and parameters (score, algorithm determination, sound pressure) are recognized. Information) and the process proceeds to step S3.

On the other hand, in step S2 as the second acquisition step, the phrase and the parameter in which the voice recognition engine 22 (smartphone 20) voice-recognizes the same voice as in step S1 is acquired via the short-range wireless communication unit 15, and the process proceeds to step S3. move on.

Next, in step S3, the parameter of the voice recognition engine 12 acquired in step S1 is compared with the parameter of the voice recognition engine 22 acquired in step S2, and the process proceeds to step S4.

Next, in step S4, it is determined whether or not the score and the sound pressure information among the parameters output by the voice recognition engine 12 are equal to or more than a predetermined threshold value. If the score and the sound pressure information are equal to or more than the threshold value (in the case of YES), the process proceeds to step S5. If it is less than the threshold value (NO), the process proceeds to step S6. This threshold value is set to 85 or more and a score of 92 or more when the maximum value of the sound pressure information (sound pressure) is 100, for example. In the present embodiment, not only the score but also the sound pressure information is set to have a threshold value because it is more likely that accurate voice recognition can be performed when the sound pressure is larger. That is, when these conditions are satisfied, it is determined that the threshold value is exceeded. That is, the threshold value used for the determination in this step corresponds to the first threshold value.

Next, in step S5, since it is determined in step S4 that the threshold value is equal to or more than the threshold value, the phrase which is the recognition result of the voice recognition engine 12 is determined (selected) as the determination word, that is, the operation command to be executed in the subsequent step. Proceed to S19.

In step S6, of the parameters acquired from the smartphone 20, it is determined whether or not the score and the sound pressure information are equal to or more than a predetermined threshold value. If the score and the sound pressure information are equal to or more than the threshold value (YES), the process proceeds to step S7, and is less than the threshold value. If (NO), the process proceeds to step S11. This threshold value is set to, for example, 82 or more and a score of 96 or more when the maximum value of the sound pressure information (sound pressure) is 100. That is, when these conditions are satisfied, it is determined that the threshold value is exceeded. The threshold value determined in this step and the threshold value determined in step S4 may be the same value. Further, these thresholds may be appropriately set based on the installation position, the algorithm of each voice recognition engine, and the like. That is, the threshold value used for the determination in this step corresponds to the second threshold value.

Next, in step S7, since it is determined in step S6 that the value is equal to or larger than the threshold value, the phrase that is the recognition result acquired from the smartphone 20 is determined (selected) as the determination word, and the process proceeds to step S8.

Next, in step S8, it is determined whether or not the determination word determined in step S7 is in the navigation command. If there is (YES), the phrase is determined (selected) as the determination word, and the process proceeds to step S19. If not present (NO), the process proceeds to step S9. The navigation command indicates a predetermined command group used for operating the car navigation system 10. That is, in this step, it is determined whether the determined determination word is a navigation command.

Next, in step S9, since it is determined in step S8 that the navigation command does not include the determination word, the smartphone 20 is caused to perform the cooperative operation, and the process proceeds to step S10. The cooperative operation means, for example, if the smartphone 20 uses a phrase (judgment word) output by the voice recognition engine 22 and an application for internet search or navigation is installed, location information (store name or location) related to the search result. Or latitude/longitude information). This cooperative operation may be transmitted from the car navigation system 10 (CPU 13) to the smartphone 20, for example, a command instructing execution, or the smartphone 20 may perform the cooperative operation in advance subsequent to the voice recognition operation.

Next, in step S10, the result of the cooperation operation performed by the smartphone 20 in step S9 is acquired via the short-range wireless communication unit 15, and the process proceeds to step S19. That is, this cooperation operation result corresponds to the processing result information.

In step S11, since both steps S4 and S6 are less than or equal to the threshold value, that is, less than the first threshold value and less than the second threshold value, the evaluation is performed using the calculation formulas (1) and (2) below. And proceeds to step S12.
(Sound pressure x a) x ((score + judgment) x b) (1)
(Sound pressure x c) x ((score + judgment) x d) (2)

The expression (1) is an expression calculated by substituting the parameters acquired from the smartphone 20, and the expression (2) is an expression calculated by substituting the parameters output by the voice recognition engine 12. Further, the judgment is a numerical value of the algorithm judgment, and a, b, c, and d are coefficients for weighting terms by which they are multiplied. That is, a predetermined calculation is performed by weighting each of the first voice recognition processing information and the second voice recognition processing information.

Next, in step S12, as a result of the evaluation performed in step S11, it is determined whether or not it can be determined as one. If it can be determined (in the case of YES), the process proceeds to step S13, and it cannot be determined. If (NO), the process proceeds to step S14. In this step, for example, when the difference between the calculation results of the above formulas (1) and (2) is 8 or more, a phrase having many calculation results is selected and fixed as one.

Next, in step S13, the phrase confirmed in step S12 or steps S15 and S17 described later is selected as a determination word, and the process proceeds to step S19.

In step S14, evaluation is performed based on the past usage history of the phrase acquired as the recognition result, and the process proceeds to step S15. The past usage history is not limited to, for example, the history of voice recognition, but is the history information used by the operation of the car navigation system 10 such as the Internet search or the destination search and the smartphone 20. Note that the usage history information of the smartphone 20 may be acquired by designating a phrase via the short-range wireless communication unit 15 when executing this step, for example.

Next, in step S15, as a result of the evaluation performed in step S14, it is determined whether or not it can be confirmed as one. If it can be confirmed (in the case of YES), the process proceeds to step S13, and it cannot be confirmed. If (NO), the process proceeds to step S16. In this step, for example, as a result of step S14, a phrase that is frequently used is selected and fixed as one.

Next, in step S16, evaluation is performed based on the past usage of the phrase acquired as the recognition result, and the process proceeds to step S15. The past usage status is a scene in which the phrase has been used in the past, for example, a time zone such as am/pm, the season, and the external environment such as weather.

Next, in step S17, as a result of the evaluation performed in step S16, it is determined whether or not it can be confirmed as one. If it can be confirmed (in the case of YES), the process proceeds to step S13 and it cannot be confirmed. If (NO), the process proceeds to step S18. In this step, for example, as a result of step S16, phrases used in the same scene are selected and confirmed as one.

Next, in step S18, the phrase with the highest score is determined as the determination word because one could not be determined in steps S12 to S17, and the process proceeds to step S19.

Next, in step S19, the voice recognition engines 12 and 22 are made to learn the determination words determined in steps S5, S7, S13, and S18, and the process proceeds to step S20. Since this learning is performed not only by the voice recognition engine 12 but also by the voice recognition engine 22, the information of the determination word is also transmitted to the smartphone 20 via the short-range wireless communication unit 15.

Next, in step S20, the command is executed based on the determination word. That is, the determination word is interpreted as an operation command of the car navigation system 10 and the process is executed. In addition, when the short-range wireless communication unit 15 acquires the cooperation result from the smartphone 20 by executing step S10, the location search may be performed based on the result, or the content may be displayed as it is. .. Further, when the determination word cannot be interpreted as an operation command, an error message may be displayed on the display unit 17, or another input may be prompted to restart the flowchart from the beginning.

As is clear from the above description, in steps S4 to S20, the phrase output by the voice recognition engine 12 or the phrase output by the voice recognition engine 22 is determined based on the parameter acquired in step S1 and the parameter acquired in step S2. It functions as a control process for selecting either one and causing the processing unit to execute information processing based on the selected phrase.

The operations shown in steps S11, S14, and S16 are not limited to this order. Alternatively, only one or two of these three operations may be performed instead of performing all of them.

Here, a specific example will be described with reference to the flowchart shown in FIG. For example, when the user pronounces “I like soba (soba),” the voice recognition engine 12 of the car navigation system 10 determines that the sound pressure information is 87 and is “Soma city”, and “Mobara city” as another candidate. It is assumed that the voice recognition engine 22 of the smartphone 20 determines that the sound pressure information is “on the sofa” when the sound pressure information is 78, and lists “soba lover” and “Soma city” as other candidates. Then, the score including each of the other candidates and the algorithm determination are as shown in FIG. As shown in FIG. 4, the flow chart shown in FIG. 2 is not limited to the comparison of one phrase for each of the car navigation system 10 and the smartphone 20, but may be comparison for a plurality of candidates.

At this time, in step S4, “Soma city” determined by the voice recognition engine 12 and “Mobara city” which is another candidate cannot satisfy the thresholds of the sound pressure information of 85 or more and the score of 92 or more. Therefore, step S6 is executed, but the "sofa" determined by the voice recognition engine 22 and the other candidates "soba lover" and "Soma city" also satisfy the thresholds of sound pressure information 82 or more and score 96 or more. I can't.

Therefore, the equations (1) and (2) are calculated and evaluated in step S11, and the determination is made in step S12. At this time, the algorithm judgment is calculated by converting points such as ⊚ and ○ into appropriate points. As a result of the calculation, for example, "on the sofa" is 78, "I like soba" is 76, "Soma-shi" is 73, and "Mobara-shi" is 41. Then, "on the sofa", "I like soba", and "Soma city" are extracted as the candidate with the highest score and candidates within 8 points from the candidate, but they cannot be determined as one. It should be noted that "Soma City" is listed as a candidate for both the car navigation system 10 and the smartphone 20, but in the subsequent determination, the one with a large calculation result of the above formula, for example, the result of the car navigation system 10 Judgment shall be made based on this. Alternatively, the voice recognition engine 12 provided with the CPU 13 that executes this flowchart may be prioritized.

Then, in step S14, the car navigation system 10 and the smartphone 20 are evaluated based on the past usage history, and the determination is made in step S15. The use history (number of times of use) of “on the sofa”, “I like soba”, and “Soma city” is as shown in FIG. Here, the items having the history in both the car navigation system 10 and the smartphone 20 (the number of times is 1 or more) are extracted. In this case, “Soba likes” and “Soma city” are extracted, but they cannot be determined as one. It should be noted that both have no history and may be narrowed down by the number of times or more and the difference from the maximum number of times or more. Further, when all the candidates are 0 times, the following calculation (step S16) is performed for all the candidates (“on the sofa”, “I like soba”, and “Soma city”).

Then, in step S16, evaluation is performed based on the situation in which the phrase was used in the past, and determination is made in step S17. As described above, the most appropriate one is selected based on the situation (status) such as time zone, season, and weather. That is, a similar one is selected from the situation used in the past and the situation this time. The similar determination may be performed by, for example, determining that two or more of the three statuses are the same and that the two are similar.

As a result of step S17, if the past usage of “I like buckwheat noodles” is similar to this time, “I like buckwheat noodles” is selected, and the determination word is determined in step S13. If neither “Soba likes” nor “Soma city” are similar, step S18 is executed and “Soba likes” having the highest score is selected.

Next, a case where "I like soba" satisfies the threshold value in step S6 will be described. In this case, "I like soba" is determined as the determination word in step S7, and it is determined in step S8 whether or not it is in the navigation command. Since "I like soba" is not included in the commands for operating the car navigation system 10, the cooperative operation of the smartphone 20 is performed in step S9. On the smartphone 20, a search related to "Soba lover" is performed using the Internet or the like, and, for example, if information on the name and location of a restaurant, a famous store, or a soba making experience event is obtained, those information are used as the cooperation result. The car navigation system 10 acquires by the smartphone 20 transmitting (step S10).

The information obtained in step S10 is used in step S20 after learning the determination word (“I like soba”) in step S19. For example, it is registered as point information based on the information such as the name and location of the restaurant, the soba-making experience event place, etc. obtained in step S10, inquired whether it is set as a destination, or displayed on a map. That is, in this case, the command execution in step S20 does not mean interpreting the determination word as an operation command, but selecting and executing an arbitrary command based on the obtained information.

According to the present embodiment, the CPU 13 of the car navigation system 10 obtains the phrase recognized by the voice recognition engine 12 that recognizes the voice uttered by the user and the parameters including the sound pressure information, the score, and the algorithm determination. Further, the voice recognition engine 22 of the smartphone 20 that has recognized the same voice and acquires a phrase and a parameter including sound pressure information, a score, and an algorithm determination via the short-range wireless communication unit 15. Then, either the recognition result of the voice recognition engine 12 or the recognition result of the voice recognition engine 22 is selected and executed as a command based on the parameters of the voice recognition engine 12 and the parameters of the voice recognition engine 22. By doing so, since it is possible to select from the results of the two voice recognition engines, it is possible to perform voice recognition with accuracy higher than that of performing voice recognition independently. Moreover, since the car navigation system 10 and the smartphone 20 have different algorithms and dictionaries, it is possible to obtain recognition results suitable for various environments. Therefore, an appropriate recognition result can be obtained.

When the parameter of the voice recognition engine 12 is equal to or more than the threshold value, the phrase recognized by the car navigation system 10 (voice recognition engine 12) is selected as the determination word, and therefore the result recognized by the car navigation system 10 is prioritized. Can be used for various purposes.

If the parameter of the voice recognition engine 12 is less than the threshold value and the parameter of the smartphone 20 (voice recognition engine 22) is equal to or more than the threshold value, the phrase recognized by the smartphone 20 is selected as the determination word. The result recognized by the smartphone 20 can be used when the reliability of the result recognized by 10 is low and there is a high possibility that the result is not suitable for use.

In addition, when the parameter of the voice recognition engine 12 is less than the threshold value and the parameter of the smartphone 20 (voice recognition engine 22) is the threshold value or more, and the recognition result of the smartphone 20 is not in the navigation command, the smartphone 20 performs the cooperative operation. The processing is performed, the result is acquired, and the processing is performed in the CPU 13. By doing so, the reliability of the result recognized by the smartphone 20 is high, but when the phrase is not a command for operating the car navigation system 10, the information related to the phrase is obtained and operated. You can

Further, since the car navigation system 10 and the smartphone 20 are made to learn the determined determination word after the determination word is determined, both speech recognition engines can share the recognition result to improve the accuracy of the subsequent speech recognition. You can In this case, in the car navigation system 10, it is possible to learn a result that has not been known so far, and it is expected that the accuracy of the score from the next time onward will be improved and the vocabulary of the dictionary will be increased. For example, new words and buzzwords can be learned in a timely manner. Further, since the navigation command can be learned in the smartphone 20, it is possible to further improve the voice recognition accuracy.

When the parameter of the voice recognition engine 12 is less than the threshold value and the parameter of the voice recognition engine 22 is less than the threshold value, evaluation by the expressions (1) and (2), evaluation by the past usage history, past usage information Since the determination word is determined based on the evaluation by etc., the determination word can be determined even when the determination word cannot be determined based on the sound pressure information, the score, and the result of the algorithm determination.

Further, since the voice recognition engine 12, the CPU 13, and the short-range wireless communication unit 15 are integrally provided, the car navigation system 10 is appropriately recognized by linking with the smartphone 20 having the voice recognition engine 22. The result can be obtained.

In the flowchart shown in FIG. 3, after the recognition result of the smartphone 20 is used as the determination word in step S7, it is determined in step S8 whether or not it is a navigation command. However, such determination is not performed, and in step S7. If determined as the determination word, step S19 may be directly executed.

Further, in the configuration shown in FIGS. 1 and 2, the car navigation system 10 and the smartphone 20 have voices input to the microphones 11 and 21, respectively. For example, the voice input to the microphone 11 of the car navigation system 10 is output as voice. The signal may be converted to a signal and then transmitted to the smartphone 20, and the smartphone 20 may perform voice recognition based on the voice signal. In this case, the sound pressure information cannot be used as a parameter, but since the score and the algorithm determination are often different, it is possible to make the determination using only this information. That is, the number of input units may be one.

Further, by configuring the flowchart of FIG. 3 as a computer-executable program, it can be configured as a control program of the information control device.

Further, although the car navigation system 10 mainly operates in the above-described embodiment, the smartphone 20 may mainly operate. Further, the information processing device is not limited to the car navigation system 10 and the smartphone 20, and may be applied to other information processing devices such as personal computers and voice-operable home electric appliances.

The present invention is not limited to the above embodiment. That is, those skilled in the art can carry out various modifications according to the conventionally known knowledge without departing from the gist of the present invention. As long as such a modification still has the configuration of the information processing apparatus of the present invention, it is of course included in the scope of the present invention.

10 Car navigation system (information processing device)
11 Microphone (input section)
12 Speech recognition engine (first speech recognition unit)
13 CPU (control unit, first acquisition unit, processing unit)
15 Short-distance wireless communication unit (second acquisition unit)
20 Smartphone 21 Microphone (input part)
22 Speech recognition engine (second speech recognition unit)
S1 Navi side voice recognition (first acquisition step)
S2 Speech recognition on the smartphone side (second acquisition step)
S4~S20 Select either one of navigation or smartphone and execute command (control process)

Claims (1)

A first acquisition unit that acquires first voice recognition result information and first voice recognition processing information from a first voice recognition unit that recognizes voice;
A second acquisition unit that acquires second voice recognition result information and second voice recognition processing information from the second voice recognition unit that recognizes the voice;
Either the first voice recognition result information or the second voice recognition result information is selected based on the first voice recognition process information and the second voice recognition process information, and the selected first voice recognition result information or A control unit that causes a processing unit to perform processing relating to the second voice recognition result information;
An information processing apparatus comprising:

Images