JP2008309864A - Voice recognition device and voice recognition method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice recognition device and a voice recognition method, in which a speaker who speaks to an in-vehicle device for controlling the device is free from switch press operation, and which causes no malfunction by clearly recognizing whether or not the speaking is to the in-vehicle device. <P>SOLUTION: In the voice recognition device 10a, when a voice recognition processing result determination processing section 13b determines that an uttered vocabulary received by a voice recognition processing section 13a is included in a keyword dictionary 12a, the voice recognition processing section 13a converts it to a corresponding command by referring to a voice recognition dictionary 12b, and forwards the voice recognition result to a command conversion output processing section 13c for outputting to a car navigation device 20. When the voice recognition processing result determination processing section 13b determines that the uttered vocabulary received by the voice recognition processing section 13a is not included in the keyword dictionary 12a, the voice recognition processing section 13a does not forward the voice recognition result to the command conversion output processing section 13c. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention provides a speech recognition means for recognizing a speech vocabulary spoken by a passenger of a vehicle, and a command for converting the speech vocabulary recognized by the speech recognition means into a corresponding command and delivering it to an in-vehicle device In particular, the speech recognition device having the conversion means and the speech recognition method using the speech recognition device free from the trouble of pressing down the talk switch for the speaker who speaks for the control toward the in-vehicle device, and The present invention relates to a speech recognition device and a speech recognition method that clearly recognizes whether or not an utterance is directed to an in-vehicle device and does not cause a malfunction.

  In recent years, various ideas have been made for realizing a technology for recognizing a user's voice. If the user's voice can be recognized, it is possible for the user to perform various device operations by voice. Especially, in-vehicle devices are concerned about the influence of manual operation by the driver on the driving. The practical application of operation technology is eagerly desired.

  By the way, at present, an in-vehicle device equipped with a vehicle voice recognition function has a specification for recognizing a specific command uttered after pressing a talk switch. By using this talk switch, the in-vehicle device can recognize a specific command more accurately (see, for example, Patent Document 1).

JP-A-10-97281

  However, in the conventional technique represented by the above-mentioned Patent Document 1, the operation of pressing the talk switch is a burden on the speaker. In particular, when the speaker is a driver, a load other than the driving operation is given. It becomes a factor. Therefore, in the future, it is expected that continuous speech recognition without using a talk switch will become the mainstream.

  However, even when a speaker on the vehicle speaks to a device other than the vehicle-mounted device (for example, utterance or self-speaking toward the passenger), the vehicle-mounted device recognizes the utterance to the device itself, There was a risk of malfunctioning based on this utterance.

  The present invention has been made to solve the above problems (problems), and frees the speaker who speaks to the in-vehicle device for the control from the trouble of pressing the talk switch, and It is an object of the present invention to provide a voice recognition device and a voice recognition method that clearly recognizes whether or not the utterance is for an in-vehicle device and does not cause a malfunction.

  In order to solve the above-described problems and achieve the object, the present invention provides a speech recognition unit that recognizes a speech vocabulary spoken by a vehicle occupant and a speech vocabulary that is speech-recognized by the speech recognition unit. Whether or not the utterance vocabulary recognized by the voice recognition means is an utterance directed to the in-vehicle device. Voice recognition result determination means for determining whether the utterance vocabulary recognized by the voice recognition means is an utterance directed to the in-vehicle device. The speech vocabulary recognized by the speech is transferred to the command conversion means only when the command is received.

  Further, the present invention provides the speech recognition unit according to the above invention, wherein when the speech recognition result determination unit determines that the utterance vocabulary recognized by the speech recognition unit is a specific vocabulary, the speech recognition unit The speech vocabulary that has been voice-recognized is transferred to the command conversion means.

  Further, the present invention provides the speech recognition unit according to the above invention, wherein when the speech recognition result determination unit determines that the utterance vocabulary speech-recognized by the speech recognition unit is a specific vocabulary, The speech vocabulary that has been voice-recognized is transferred to the command conversion means.

  Further, the present invention provides the speech recognition unit according to the above invention, wherein the speech recognition result determination unit determines that the utterance vocabulary recognized by the speech recognition unit is the first specific vocabulary. An utterance vocabulary speech-recognized after the first specific vocabulary is started to be transferred to the command conversion means, and an utterance vocabulary speech-recognized after the first specific vocabulary by the speech recognition means is a second specific vocabulary. When the speech recognition result determination means determines that the vocabulary is a vocabulary, the speech vocabulary obtained by speech recognition of the utterance vocabulary speech recognized after the second specific vocabulary is terminated to the command conversion means. It is characterized by.

  Further, according to the present invention, in the above invention, buffering means for buffering a predetermined number of utterance vocabulary speech-recognized by the speech recognition means, and a category to which the utterance vocabulary belongs is stored in association with the utterance vocabulary. Vocabulary category storage means, and the speech recognition means is an utterance in which the utterance vocabulary recognized by speech based on the category of the utterance vocabulary buffered in the buffering means is directed to the in-vehicle device. The speech vocabulary that has been speech-recognized is transferred to the command conversion means only when it is determined by the speech recognition result determination means.

  Further, the present invention provides the speech recognition unit according to the above-described invention, wherein the speech recognition unit assumes that the appearance rate of the specific category is a predetermined value or more in the speech vocabulary buffered by the buffering unit. When the speech recognition result determining means determines that the speech is directed to the in-vehicle device, the speech vocabulary recognized by the speech is transferred to the command conversion means.

  Further, the present invention provides the speech recognition unit according to the above-described invention, wherein the speech recognition unit assumes that the appearance rate of the specific category is a predetermined value or more in the speech vocabulary buffered by the buffering unit. Canceling the delivery of the speech-recognized speech vocabulary to the command conversion means after a predetermined number of times that the speech recognition result judgment means has not determined that the speech is directed to the in-vehicle device It is characterized by doing.

  Further, according to the present invention, in the above invention, the voice recognition unit assumes that a specific category continues a predetermined number of times in the utterance vocabulary buffered by the buffering unit, and the utterance vocabulary that has been voice-recognized is transferred to the in-vehicle device. When the speech recognition result determining unit determines that the utterance is directed, the speech vocabulary recognized by the speech is transferred to the command conversion unit.

  Further, according to the present invention, in the above invention, the voice recognition unit assumes that a specific category continues a predetermined number of times in the utterance vocabulary buffered by the buffering unit, and the utterance vocabulary that has been voice-recognized is transferred to the in-vehicle device. Canceling the delivery of the speech-recognized speech vocabulary to the command conversion means after a predetermined number of times that the speech recognition result judgment means has not determined that the speech is directed. To do.

  The present invention also provides a speech recognition step for recognizing speech vocabulary spoken by a vehicle occupant and converting the speech vocabulary speech recognized by the speech recognition means into a corresponding command to be received by an in-vehicle device. A speech recognition method including a command conversion step to pass, further including a speech recognition result determination step of determining whether or not the utterance vocabulary recognized by the speech recognition step is an utterance directed to the in-vehicle device. In the speech recognition step, only when the speech recognition result determination step determines that the utterance vocabulary recognized by the speech recognition step is an utterance directed to the in-vehicle device, the speech recognition vocabulary recognized by the speech Is transferred to the affirmative command conversion.

  According to the present invention, the speech recognition means converts the speech utterance vocabulary recognized by the speech recognition result determination means into a command conversion means only when the speech recognition result determination means determines that the speech utterance vocabulary is directed to the in-vehicle device. Since the talk switch is not required, the passenger is freed from the troublesome operation of pressing the talk switch, and the voice vocabulary is directed to the in-vehicle device by voice recognition. It is possible to always distinguish the utterance vocabulary that is not a thing and prevent the in-vehicle device from malfunctioning due to a command based on erroneous voice recognition.

  Further, according to the present invention, when the speech recognition result determining means determines that the utterance vocabulary recognized by the speech recognition means is the specific vocabulary, the command conversion is performed on the utterance vocabulary recognized after the specific vocabulary. Therefore, the voice recognition means clearly recognizes the start of the utterance without the need to press the talk switch for starting the utterance for controlling the in-vehicle device. It has the effect of being free from bothersomeness.

  Further, according to the present invention, when the speech recognition result determining means determines that the utterance vocabulary recognized by the speech recognition means is the specific vocabulary, the command conversion is performed on the utterance vocabulary recognized after the specific vocabulary. Therefore, the voice recognition means clearly recognizes the start of the utterance without the need to press the talk switch for starting the utterance for controlling the in-vehicle device, and ends the utterance for controlling the in-vehicle device. Can be clearly recognized by the voice recognition means, and the passenger is freed from the troublesome operation of pressing the talk switch.

  Further, according to the present invention, the speech recognition means determines the speech recognition result based on the category of the utterance vocabulary buffered by the buffering means if the utterance vocabulary recognized by the speech is directed to the in-vehicle device. Since the speech vocabulary recognized by the voice is transferred to the command conversion means only when it is determined by the means, the passenger does not have to be aware of the start of the utterance for controlling the in-vehicle device while always performing the voice recognition. In addition, the vocabulary directed to the in-vehicle device is distinguished from the utterance vocabulary that is not directed to the in-vehicle device, reducing the burden on the speaker and causing the in-vehicle device to malfunction due to an incorrect voice recognition command. This has the effect of preventing this.

  Further, according to the present invention, the speech recognition means determines that the appearance rate of the specific category among the utterance vocabulary buffered by the buffering means exceeds a predetermined value, and the speech utterance vocabulary is transferred to the in-vehicle device. When the speech recognition result determining means determines that the utterance is directed, the speech recognition speech vocabulary is transferred to the command conversion means, so that by recognizing that the utterance content shows a specific tendency, Even if the passenger is not conscious of the utterance for controlling the in-vehicle device, the utterance vocabulary directed to the in-vehicle device and the utterance vocabulary not directed to the in-vehicle device are distinguished, reducing the burden on the utterer At the same time, the in-vehicle device is prevented from malfunctioning due to a command based on erroneous voice recognition.

  Further, according to the present invention, the speech recognition means determines that the appearance rate of the specific category among the utterance vocabulary buffered by the buffering means exceeds a predetermined value, and the speech utterance vocabulary is transferred to the in-vehicle device. Since the case where the speech recognition result determination means does not determine that the speech is directed is canceled after a predetermined number of consecutive times, the speech utterance vocabulary transferred to the command conversion means is canceled. By recognizing that a specific tendency is no longer shown, utterance vocabulary intended for in-vehicle devices and in-vehicle devices, even if the passenger is not aware of the end of the utterance to control the in-vehicle devices Utterance vocabulary is not distinguished, reducing the burden on the speaker, and preventing the in-vehicle device from malfunctioning due to commands based on incorrect voice recognition. An effect.

  Further, according to the present invention, the speech recognition result indicates that the speech utterance vocabulary is an utterance directed to the in-vehicle device, assuming that the specific category continues for a predetermined number of times in the utterance vocabulary buffered by the buffering means. When judged by the judging means, the voice-recognized utterance vocabulary is transferred to the command converting means, so that the passenger can recognize the utterance content strongly even if it shows a specific tendency temporarily. Even if you are not conscious of the utterances to control the in-vehicle device, the utterance vocabulary directed to the in-vehicle device and the utterance vocabulary not directed to the in-vehicle device are distinguished, reducing the burden on the speaker and This has the effect of preventing the in-vehicle device from malfunctioning due to a command based on the voice recognition.

  Further, according to the present invention, the speech recognition means determines that a specific category continues for a predetermined number of times in the utterance vocabulary buffered by the buffering means, and the speech recognition utterance vocabulary is directed to the in-vehicle device. If the speech recognition result determination means does not determine that the speech recognition result is determined, the speech utterance vocabulary is cancelled from being transferred to the command conversion means after a predetermined number of consecutive times. By recognizing that the utterance is no longer shown strongly, the utterance vocabulary for the in-vehicle device and the in-vehicle device are not intended even if the passenger is unaware of the end of the utterance for controlling the in-vehicle device. It is distinguished from the utterance vocabulary not to reduce the burden on the speaker and to prevent the in-vehicle device from malfunctioning due to a command based on erroneous voice recognition. Achieve the cormorant effect.

  Exemplary embodiments according to a speech recognition apparatus and speech recognition method of the present invention will be described below in detail with reference to the accompanying drawings.

  Embodiment 1 according to the present invention will be described below with reference to FIGS. In the first embodiment, when a specific keyword set in advance is recognized by voice recognition in the continuous voice recognition of an utterance vocabulary by a vehicle occupant, the utterance vocabulary recognized immediately after the keyword is converted into a car navigation device or the like. This is an embodiment in which a command is transferred to a predetermined command conversion unit in order to convert it into a controllable command.

  First, the configuration of the speech recognition apparatus according to the first embodiment will be described. FIG. 1 is a functional block diagram of the configuration of the speech recognition apparatus according to the first embodiment. As shown in the figure, in a vehicle 1, via a CAN (Controller Area Network) 2, a voice recognition device 10a according to the first embodiment and a car navigation that is a control target by a control command based on the speech-recognized utterance content. The apparatus 20 is connected. In the following embodiment, a control command for controlling an in-vehicle device such as the car navigation device 20 in which the speech-recognized utterance content is converted is simply referred to as a “command”.

  The voice recognition device 10a includes a display unit 11a that is a display unit such as a display device having a predetermined display screen, a voice generation unit 11b that is a voice generation unit such as a speaker device that emits voice, and a volatile or nonvolatile memory. It has the memory | storage part 12 and the control part 13 which are means. The voice recognition device 10a is connected to a microphone 14 for inputting voice data detected from the outside to the voice recognition device 10a.

  The storage unit 12 stores a keyword dictionary 12a and a voice recognition dictionary 12b. The keyword dictionary 12a and the speech recognition dictionary 12b are stored in the storage unit 12 as predetermined tables. The keyword dictionary 12a is a list of specific words set in advance. The voice recognition dictionary 12b is a list of commands to be converted from the speech-recognized utterance contents.

  The control unit 13 is responsible for overall control of the voice recognition device 10a. In particular, the characteristic functional configuration related to the first embodiment includes a voice recognition processing unit 13a, a voice recognition processing result determination processing unit 13b, and a command conversion. And an output processing unit 13c. Other functional configurations are omitted.

  When the vocabulary spoken by the passenger of the vehicle 1 is detected by the microphone 14, the speech recognition processing unit 13 a accepts the detected utterance vocabulary for a while and temporarily stores it. When the utterance vocabulary is included in the keyword dictionary 12a (the above is called speech recognition in the previous stage), the vocabulary detected after the utterance vocabulary is subjected to speech recognition processing (speech recognition by this speech recognition processing is performed) Called later speech recognition). Note that the term “speech recognition” simply indicates a case where it is determined that the detected utterance vocabulary is included in the speech recognition dictionary 12b, and the speech recognition process is a process of attempting speech recognition.

  The speech recognition processing result determination processing unit 13b is a processing unit that determines whether or not the utterance vocabulary accepted by the speech recognition processing unit 13a is included in the keyword dictionary 12a. When the speech recognition processing result determination processing unit 13b determines that the utterance vocabulary accepted by the speech recognition processing unit 13a is included in the keyword dictionary 12a, the speech recognition processing unit 13a outputs the speech recognition result as a command conversion output. The data is transferred to the processing unit 13c. When the speech recognition processing result determination processing unit 13b does not determine that the utterance vocabulary accepted by the speech recognition processing unit 13a is included in the keyword dictionary 12a, the speech recognition processing unit 13a uses the speech recognition result as a command. The data is not delivered to the conversion output processing unit 13c.

  The command conversion output processing unit 13c converts the voice recognition result delivered from the voice recognition processing unit 13a into a corresponding command with reference to the voice recognition dictionary 12b, and outputs the command to the car navigation device 20.

  Next, the speech recognition process (part 1) of the first embodiment will be described. The voice recognition process (No. 1) according to the first embodiment is a process in the case of converting a vocabulary speech-recognized immediately after the keyword into a command when the voice recognition result by the voice recognition in the previous stage is a keyword. FIG. 2 is a flowchart illustrating the speech recognition processing procedure (part 1) according to the first embodiment.

  First, the speech recognition processing unit 13a performs speech recognition processing on the utterance vocabulary input via the microphone 14, and outputs the processing result to the speech recognition processing result determination processing unit 13b (step S101). Subsequently, the speech recognition processing result determination processing unit 13b determines whether or not the “command conversion flag” stored in the predetermined storage area is on (step S102). When it is determined that the “command conversion flag” is ON (Yes at Step S102), the process proceeds to Step S103, and when it is not determined that the “command conversion flag” is ON (No at Step S102), Step S105 is performed. Move on.

  In step S103, the speech recognition processing unit 13a delivers the speech recognition processing result of the input utterance vocabulary to the command conversion output processing unit 13c based on the determination processing result by the speech recognition processing result determination processing unit 13b. Subsequently, the speech recognition processing result determination processing unit 13b turns off the “command conversion flag” stored in the predetermined storage area (step S104).

  In step S105, the speech recognition processing result determination processing unit 13b refers to the keyword dictionary 12a and determines whether or not the speech recognition processing result input from the speech recognition processing unit 13a is a keyword. When it is determined that the speech recognition processing result input from the speech recognition processing unit 13a is a keyword (Yes in step S105), the process proceeds to step S106, and the speech recognition processing result input from the speech recognition processing unit 13a is a keyword. If it is not determined that there is any (No in step S105), the process proceeds to step S107.

  In step S106, the speech recognition processing result determination processing unit 13b turns on a “command conversion flag” stored in a predetermined storage area. In step S107, the speech recognition processing result determination processing unit 13b turns off the “command conversion flag” stored in the predetermined storage area. When these processes are completed, the process proceeds to step S108.

  In step S108, the voice recognition processing unit 13a determines whether or not to end the output of the voice recognition result to the command conversion output processing unit 13c, and ends the output of the voice recognition result to the command conversion output processing unit 13c. In the case (Yes at Step S108), the voice recognition process (No. 1) of the first embodiment is finished, and when the output of the voice recognition result to the command conversion output processing unit 13c is not finished (No at Step S108), the process goes to Step S101. Move.

  Next, the voice recognition process (part 2) of the first embodiment will be described. The voice recognition process (No. 2) of the first embodiment is a process in the case of converting a vocabulary voice-recognized immediately before the keyword into a command when the voice recognition result by the voice recognition in the previous stage is a keyword. FIG. 3 is a flowchart illustrating the voice recognition processing procedure (part 2) according to the first embodiment.

  First, the speech recognition processing unit 13a clears the “predetermined buffer” (step S111). This buffer is provided in volatile or non-volatile storage means.

  Subsequently, the speech recognition processing unit 13a performs speech recognition processing on the utterance vocabulary input via the microphone 14, and outputs the processing result to the speech recognition processing result determination processing unit 13b (step S112). Subsequently, the speech recognition processing unit 13a determines whether or not the “predetermined buffer” has been cleared (step S113). When it is determined that the “predetermined buffer” is cleared (Yes at Step S113), the process proceeds to Step S117, and when it is not determined that the “predetermined buffer” is cleared (No at Step S113), Step S114 is performed. Move on.

  In step S114, the speech recognition processing result determination processing unit 13b refers to the keyword dictionary 12a and determines whether or not the speech recognition result input from the speech recognition processing unit 13a is a keyword. When it is determined that the speech recognition result input from the speech recognition processing unit 13a is a keyword (Yes in step S114), the process proceeds to step S115, and the speech recognition result input from the speech recognition processing unit 13a is a keyword. If not determined (No at step S114), the process proceeds to step S117.

  In step S115, the speech recognition processing unit 13a delivers the speech recognition processing result of the input utterance vocabulary to the command conversion output processing unit 13c based on the determination processing result by the speech recognition processing result determination processing unit 13b. Subsequently, the speech recognition processing result determination processing unit 13b turns off the “command conversion flag” stored in the predetermined storage area (step S116).

  Subsequently, the speech recognition processing unit 13a stores the speech recognition result input in step S112 in the “predetermined buffer” (step S117). If a speech recognition result is already stored in the “predetermined buffer” during the process of step S117, the old speech recognition result is deleted and a new speech recognition result is stored. Subsequently, the voice recognition processing unit 13a determines whether or not to end the output of the voice recognition result to the command conversion output processing unit 13c (step S118), and outputs the voice recognition result to the command conversion output processing unit 13c. Is terminated (Yes at step S118), the speech recognition process (part 2) of the first embodiment is terminated, and when the output of the speech recognition result to the command conversion output processing unit 13c is not terminated (No at step S118), Control goes to step S112.

  A second embodiment according to the present invention will be described below with reference to FIGS. 4 and 5. In the first embodiment, when a specific start keyword set in advance is recognized by voice recognition in the continuous speech recognition of the utterance vocabulary by the vehicle occupant, the utterance vocabulary recognized after the keyword is converted into the car navigation device or the like. Is converted to a controllable command to a predetermined command conversion unit, and when a specific end keyword is recognized after speech recognition of a specific start keyword, a voice-recognized utterance vocabulary is It is an Example which complete | finishes delivery to a command conversion part. Only the differences from the first embodiment will be described below.

  First, the configuration of the speech recognition apparatus according to the second embodiment will be described. FIG. 4 is a functional block diagram of the configuration of the speech recognition apparatus according to the second embodiment. Compared with the speech recognition device 10a of the first embodiment, the speech recognition device 10b of the second embodiment has a configuration in which the storage unit 12 includes a start keyword dictionary 12c and an end keyword dictionary 12d instead of the keyword dictionary 12a. . Except for these configurations, the speech recognition device 10b according to the second embodiment is the same as the speech recognition device 10a according to the first embodiment, and a description thereof will be omitted.

  The start keyword dictionary 12c and the end keyword dictionary 12d are stored in the storage unit 12 as predetermined tables. The start keyword dictionary 12c and the end keyword dictionary 12d are lists of specific vocabularies set in advance. In particular, the start keyword dictionary 12c stores a keyword indicating the start of an utterance toward an in-vehicle device such as the car navigation device 20, and the end keyword dictionary 12d is directed to the in-vehicle device such as the car navigation device 20. A keyword indicating the end of the utterance is stored.

  Next, the speech recognition process according to the second embodiment will be described. In the voice recognition process of the second embodiment, when the voice recognition result by the voice recognition in the previous stage is a start keyword, the vocabulary recognized after the keyword is command-converted, and then the end keyword is voice-recognized. This is a process for ending the command conversion of the recognized vocabulary. FIG. 5 is a flowchart illustrating a voice recognition processing procedure according to the second embodiment.

  First, the speech recognition processing unit 13a performs speech recognition processing on the utterance vocabulary input via the microphone 14, and outputs the processing result to the speech recognition processing result determination processing unit 13b (step S121). Subsequently, the speech recognition processing result determination processing unit 13b determines whether or not the “command conversion flag” stored in the predetermined storage area is on (step S122). When it is determined that the “command conversion flag” is ON (Yes at Step S122), the process proceeds to Step S123, and when it is not determined that the “command conversion flag” is ON (No at Step S122), Step S126. Move on.

  In step S123, the speech recognition processing result determination processing unit 13b refers to the end keyword dictionary 12d and determines whether or not the speech recognition result input from the speech recognition processing unit 13a is an end keyword. When it is determined that the speech recognition result input from the speech recognition processing unit 13a is the end keyword (Yes at Step S123), the process proceeds to Step S124, and the speech recognition result input from the speech recognition processing unit 13a is the end keyword. If it is not determined that there is (No at Step S123), the process proceeds to Step S125.

  In step S124, the speech recognition processing result determination processing unit 13b turns off the “command conversion flag” stored in the predetermined storage area. In step S125, the speech recognition processing unit 13a delivers the speech recognition processing result of the input utterance vocabulary to the command conversion output processing unit 13c based on the determination processing result by the speech recognition processing result determination processing unit 13b.

  On the other hand, in step S126, the speech recognition processing result determination processing unit 13b refers to the start keyword dictionary 12c and determines whether or not the speech recognition result input from the speech recognition processing unit 13a is a start keyword. When it is determined that the speech recognition result input from the speech recognition processing unit 13a is the start keyword (Yes at Step S126), the process proceeds to Step S127, and the speech recognition result input from the speech recognition processing unit 13a is the start keyword. If it is not determined that there is any (No in step S126), the process proceeds to step S129.

  In step S127, the speech recognition processing unit 13a delivers the speech recognition processing result of the input utterance vocabulary to the command conversion output processing unit 13c based on the determination processing result by the speech recognition processing result determination processing unit 13b. Subsequently, the speech recognition processing result determination processing unit 13b turns on the “command conversion flag” stored in the predetermined storage area (step S128). When this process ends, the process proceeds to step S129.

  In step S129, the speech recognition processing unit 13a determines whether or not to end the output of the speech recognition result to the command conversion output processing unit 13c, and ends the output of the speech recognition result to the command conversion output processing unit 13c. In such a case (Yes at Step S129), the voice recognition process of the second embodiment is finished. When the output of the voice recognition result to the command conversion output processing unit 13c is not finished (No at Step S129), the process proceeds to Step S121.

  A third embodiment according to the present invention will be described below with reference to FIGS. In the third embodiment, in the continuous speech recognition of the utterance vocabulary by the vehicle occupant, the vocabulary in the specific category occupies a predetermined ratio among the vocabulary recognized in a certain time, or the vocabulary in the specific category continues for a certain number of times. When speech recognition is performed, an utterance vocabulary recognized after the above conditions are satisfied is transferred to a predetermined command conversion unit to convert the car navigation device into a controllable command. It is an example. In the following, only the differences from the first or second embodiment will be described.

  First, the configuration of the speech recognition apparatus according to the third embodiment will be described. FIG. 6 is a functional block diagram of the configuration of the speech recognition apparatus according to the third embodiment. Compared with the speech recognition device 10a of the first embodiment, the speech recognition device 10c of the third embodiment includes a vocabulary category classification table 12e and a recognized vocabulary storage buffer 12f in the storage unit 12 instead of the keyword dictionary 12a. ing. Except for these configurations, the speech recognition device 10c according to the third embodiment is the same as the speech recognition device 10a according to the first embodiment, and thus the description thereof is omitted.

  The vocabulary category classification table 12e is a table that stores at least one affiliation category associated with a speech-recognized utterance vocabulary. For example, as shown in the table example in FIG. 7, “meal”, “Japanese food”, and the like are associated with “vocabulary” and “soba” as “affiliation category”. In addition, “navigation”, “map”, and the like are associated with “vocabulary” and “destination” as “affiliation category”. “Vocabulary” “xx television station” is associated with “TV”, “audio”, etc. as “affiliation category”.

  As described above, at least one affiliation category of the vocabulary determined to be included in the vocabulary category classification table 12e by the speech recognition processing result determination processing unit 13b is acquired based on the vocabulary category classification table 12e.

  The recognized vocabulary storage buffer 12f is a storage area for buffering a predetermined number (for example, 500 vocabularies) of words that have been continuously recognized by the speech recognition processing unit 13a. The vocabulary buffered in the recognized vocabulary storage buffer 12f is managed in a first-in first-out manner, and if a vocabulary newly recognized by voice exceeding the predetermined number is stored, it is stored the oldest in terms of time. The vocabulary is deleted and the newly recognized vocabulary is stored.

  Next, the speech recognition process (No. 1) according to the third embodiment will be described. In the speech recognition process (No. 3) of the third embodiment, it is determined that the ratio of the vocabulary belonging to the same category among the vocabulary of the speech recognition result buffered in the recognized vocabulary storage buffer 12f is equal to or greater than a predetermined threshold. In this case, the vocabulary recognized after the determination is command-converted. FIG. 8 is a flowchart illustrating the speech recognition processing procedure (part 1) according to the third embodiment.

  First, the speech recognition processing unit 13a performs speech recognition processing on the utterance vocabulary input via the microphone 14, and outputs the processing result to the speech recognition processing result determination processing unit 13b (step S131). Subsequently, the speech recognition processing result determination processing unit 13b refers to the vocabulary category classification table 12e and acquires the affiliation category of the input speech recognition result vocabulary (step S132).

  Subsequently, the speech recognition processing result determination processing unit 13b buffers the input speech recognition result by a predetermined number (for example, 500 vocabulary) in the recognized vocabulary storage buffer 12f (step S133).

  Subsequently, the speech recognition processing result determination processing unit 13b determines that the ratio of the vocabulary of the same category among the vocabulary of the speech recognition result buffered in the recognition vocabulary storage buffer 12f is equal to or higher than a predetermined threshold (for example, 80%). It is determined whether or not there is (step S134). When it is determined that the ratio of the vocabulary of the same category belongs to the vocabulary of the same affiliation category among the vocabulary of the speech recognition result buffered in the recognition vocabulary storage buffer 12f (Yes in step S134), the process proceeds to step S135, and the recognition vocabulary storage buffer If it is not determined that the vocabulary of the same affiliation category among the vocabulary of the speech recognition result buffered in 12f is greater than or equal to the predetermined threshold (No at step S134), the process proceeds to step S137.

  In step S135, the speech recognition processing unit 13a transfers the speech recognition processing result of the input utterance vocabulary to the command conversion output processing unit 13c based on the determination processing result by the speech recognition processing result determination processing unit 13b. Subsequently, the speech recognition processing result determination processing unit 13b turns on the “command conversion flag” stored in a predetermined storage area (step S136). When this process ends, the process proceeds to step S141.

  On the other hand, in step S137, the speech recognition processing result determination processing unit 13b determines whether or not the “command conversion flag” stored in the predetermined storage area is on. When it is determined that the “command conversion flag” is on (Yes at step S137), the process proceeds to step S138. When it is not determined that the “command conversion flag” is on (No at step S137), step S139 is performed. Move on.

  Subsequently, the speech recognition processing result determination processing unit 13b does not determine the predetermined number of times that the vocabulary of the same affiliation category in the vocabulary of the speech recognition result buffered in the recognition vocabulary storage buffer 12f is equal to or greater than a predetermined threshold. (That is, whether negative in step S134 has continued for a predetermined number of times) or not (step S139). When step S134 is negative for a predetermined number of times (Yes at step S139), the speech recognition processing result determination processing unit 13b turns off the “command conversion flag” stored in the predetermined storage area (step S140), and negative at step S134. Does not continue for a predetermined number of times (No at step S139), the process proceeds to step S141.

  In step S141, the speech recognition processing unit 13a determines whether or not to end the output of the speech recognition result to the command conversion output processing unit 13c, and ends the output of the speech recognition result to the command conversion output processing unit 13c. In such a case (Yes at Step S141), the voice recognition process of the third embodiment is finished. When the output of the voice recognition result to the command conversion output processing unit 13c is not finished (No at Step S141), the process proceeds to Step S131.

  When the determination processing in step S134 described above shows some tendency in the content of the utterance content due to the categorization of the utterance vocabulary, the speech recognition result is transferred to the command conversion processing unit, and the command conversion is performed. When the tendency of the utterance contents due to the categorization of the utterance vocabulary does not appear as a result of the determination process of S139, control is performed so that the command conversion is not performed by transferring the speech recognition result to the command conversion processing unit. It becomes possible.

  In addition, instead of “whether or not the ratio of the vocabulary of the same affiliation category among the vocabulary of the speech recognition result buffered in the recognized vocabulary storage buffer 12f is a predetermined threshold or more” It may be adopted whether or not a predetermined number of vocabularies of the same category belong to the vocabulary of the speech recognition result buffered in the vocabulary storage buffer 12f. This is changed to step S134a in place of step S134 in FIG. 8 (see FIG. 9).

  Accordingly, the determination condition in step S139 of FIG. 8 is that “the ratio of the vocabulary of the same affiliation category in the vocabulary of the speech recognition result buffered in the recognized vocabulary storage buffer 12f is a predetermined number of times or more. Instead of whether or not determination is made (that is, whether or not the negative in step S134 is continued for a predetermined number of times), a predetermined number of vocabularies of the same category belong to the vocabulary of the speech recognition result buffered in the recognized vocabulary storage buffer 12f Whether or not the predetermined number of times is determined to be continuous (that is, whether or not negative in step S134a has been continued a predetermined number of times) is adopted. This is replaced with step S139 in FIG. 8 and is referred to as step S139a (see FIG. 9).

  When a strong tendency appears temporarily in the utterance content by the category classification of the utterance vocabulary by the determination processing in the above step S134a, the voice recognition result is transferred to the command conversion processing unit to perform command conversion, and step S139a. If there is no temporary strong tendency of the utterance content due to the categorization of the utterance vocabulary by the judgment processing of, the transfer of the speech recognition result to the command conversion processing unit is canceled and the command conversion is not performed. It becomes possible to control.

  As mentioned above, although the Example of this invention was described, this invention is not limited to this, In the range of the technical idea described in the claim, even if it implements in a various different Example, it is. It ’s good. Moreover, the effect described in the Example is not limited to this.

  In addition, among the processes described in the above embodiment, all or part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed can be performed. All or a part can be automatically performed by a known method. In addition, the processing procedure, control procedure, specific name, information including various data and parameters shown in the above embodiment can be arbitrarily changed unless otherwise specified.

  Each component of each illustrated device is functionally conceptual and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured.

  Furthermore, each or all of the processing functions performed in each device are entirely or partially a CPU (Central Processing Unit) (or a microcomputer such as an MPU (Micro Processing Unit) or MCU (Micro Controller Unit)) and It may be realized by a program that is analyzed and executed by the CPU (or a microcomputer such as MPU or MCU), or may be realized as hardware by wired logic.

  According to the present invention, in a voice recognition device, a speaker who speaks for control to an in-vehicle device is released from the troublesome operation of pressing the talk switch, and whether or not the utterance is for the in-vehicle device. This is useful when you want to clearly recognize and avoid malfunctions.

1 is a functional block diagram illustrating a configuration of a speech recognition apparatus according to Embodiment 1. FIG. 6 is a flowchart illustrating a voice recognition processing procedure (part 1) according to the first embodiment. It is a flowchart which shows the speech recognition process sequence (the 2) of Example 1. FIG. It is a functional block diagram which shows the structure of the speech recognition apparatus concerning Example 2. FIG. 10 is a flowchart illustrating a voice recognition processing procedure according to the second embodiment. FIG. 6 is a functional block diagram illustrating a configuration of a speech recognition apparatus according to a third embodiment. It is a figure which shows the example of a vocabulary category classification | category table. 12 is a flowchart illustrating a voice recognition processing procedure (No. 1) according to the third embodiment. 10 is a flowchart illustrating a voice recognition processing procedure (No. 2) according to the third embodiment.

Explanation of symbols

1 vehicle 10a speech recognition device 10b speech recognition device 10c speech recognition device 11a display unit 11b speech generation unit 12 storage unit 12a keyword dictionary 12b speech recognition dictionary 12c start keyword dictionary 12d end keyword dictionary 12e vocabulary category classification table 12f recognition vocabulary storage buffer 13 Control unit 13a Speech recognition processing unit 13b Speech recognition processing result determination processing unit 13c Command conversion output processing unit 14 Microphone 20 Car navigation device

Claims (10)

Speech recognition means for recognizing speech vocabulary spoken by a passenger of the vehicle, and command conversion means for converting the speech vocabulary recognized by the speech recognition means into a corresponding command and delivering it to the in-vehicle device A voice recognition device comprising:
Further comprising speech recognition result determination means for determining whether the utterance vocabulary recognized by the voice recognition means is an utterance directed to the in-vehicle device;
The speech recognition means determines the speech vocabulary recognized by the speech recognition only when the speech recognition result determination means determines that the utterance vocabulary recognized by the speech recognition means is an utterance directed to the in-vehicle device. A speech recognition device, wherein the speech recognition device is passed to a command conversion means.   The speech recognition means, when the speech recognition result determination means determines that the utterance vocabulary recognized by the speech recognition means is a specific vocabulary, the command conversion is performed on the utterance vocabulary recognized after the specific vocabulary. The voice recognition apparatus according to claim 1, wherein the voice recognition apparatus is transferred to a means.   The speech recognition means, when the speech recognition result determination means determines that the utterance vocabulary recognized by the speech recognition means is a specific vocabulary, converts the utterance vocabulary speech recognized before the specific vocabulary to the command conversion The voice recognition apparatus according to claim 1, wherein the voice recognition apparatus is transferred to a means.   The speech recognition means recognizes speech after the first specific vocabulary when the speech recognition result determination means determines that the utterance vocabulary recognized by the speech recognition means is the first specific vocabulary. The speech recognition result determining means starts to deliver the utterance vocabulary to the command conversion means, and the speech vocabulary recognized after the first specific vocabulary by the speech recognition means is the second specific vocabulary. 2. The method according to claim 1, wherein when the determination is performed, the utterance vocabulary obtained by speech recognition of the utterance vocabulary recognized after the second specific vocabulary is terminated to the command conversion means. Voice recognition device. Buffering means for buffering a predetermined number of speech vocabularies recognized by the voice recognition means;
Vocabulary category storage means for storing the utterance vocabulary in association with the category to which the utterance vocabulary belongs, and
The voice recognition means is determined by the voice recognition result determination means that the utterance vocabulary that has been voice-recognized is an utterance directed to the in-vehicle device based on the category of the utterance vocabulary buffered in the buffering means. 2. The speech recognition apparatus according to claim 1, wherein the speech vocabulary that has been speech-recognized is transferred to the command conversion means only when the speech recognition is performed.   The speech recognition means is an utterance where the speech utterance vocabulary recognized as speech is directed to the in-vehicle device, assuming that the appearance rate of a specific category among the utterance vocabulary buffered by the buffering means is a predetermined value or more. 6. The speech recognition apparatus according to claim 5, wherein if the speech recognition result determination means determines that there is, the speech vocabulary recognized by the speech is transferred to the command conversion means.   The speech recognition means is an utterance where the speech utterance vocabulary recognized as speech is directed to the in-vehicle device, assuming that the appearance rate of a specific category among the utterance vocabulary buffered by the buffering means is a predetermined value or more. 7. The method according to claim 6, further comprising: canceling delivery of the speech-recognized utterance vocabulary to the command conversion means after a predetermined number of consecutive cases where the speech recognition result determination means does not determine. The speech recognition apparatus according to the description.   The speech recognition means recognizes that the speech recognition is a speech directed to the in-vehicle device, assuming that a specific category continues for a predetermined number of times in the speech vocabulary buffered by the buffering means. 6. The speech recognition apparatus according to claim 5, wherein, when judged by the result judging means, the speech vocabulary recognized by the speech is transferred to the command converting means.   The speech recognition means recognizes that the speech recognition is a speech directed to the in-vehicle device, assuming that a specific category continues for a predetermined number of times in the speech vocabulary buffered by the buffering means. 9. The voice recognition apparatus according to claim 8, wherein after the case where the result is not judged by the result judgment means continues for a predetermined number of times, the delivery of the utterance vocabulary recognized by the voice to the command conversion means is cancelled. . A speech recognition step for recognizing an utterance vocabulary spoken by a vehicle occupant, and a command conversion step for converting the utterance vocabulary recognized by the speech recognition means into a corresponding command and transferring it to an in-vehicle device. A speech recognition method including:
A speech recognition result determination step of determining whether or not the utterance vocabulary recognized by the speech recognition step is an utterance directed to the in-vehicle device;
In the speech recognition step, only when the speech recognition result determination step determines that the speech vocabulary speech-recognized by the speech recognition step is an utterance directed to the in-vehicle device, the speech recognition speech vocabulary is The speech recognition method, wherein the command conversion is passed to the positive.

Images