JP2015129794A - Voice recognition apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce erroneous function execution caused by a conversation with the surroundings or a monologue of an operator.SOLUTION: It is analyzed whether or not a voice-recognized speech production phrase is finished in the form of indeclinable stop or in an imperative form of a verb (21-23), and a function corresponding to the voice-recognized speech production phrase is executed on condition that the voice-recognized speech production phrase is finished in the form of indeclinable stop or an imperative form of a verb as the result of the analysis.

Description

  The present invention relates to a speech recognition apparatus that recognizes speech phrases spoken by a user.

  As this type of device, there is one in which a user's voice is constantly input, matching / non-matching of operation commands registered in advance is always checked, and a function is executed when a matching operation command is recognized (for example, , See Patent Document 1).

JP-A-2001-51694

  However, the device described in Patent Document 1 is registered when, for example, an utterance phrase included in a conversation with a surrounding person or an utterance phrase included in a user's monologue matches an operation command registered in advance. It is erroneously determined that the operated command is spoken. In this case, there is a problem that a function corresponding to the operation command that is erroneously determined is executed against the user's will.

  The present invention has been made in view of the above problems, and an object of the present invention is to reduce erroneous execution of functions due to conversation with the surroundings or driver's monologue.

  In order to achieve the above object, the invention according to claim 1 recognizes an utterance phrase spoken by a user by voice, and the utterance phrase recognized by voice recognition by this voice recognition matches a predetermined registered phrase. A speech recognition device that performs a function corresponding to a speech phrase that has been speech-recognized, and that analyzes whether the speech phrase that has been speech-recognized ends with a verbal stop or verb command form; Execute the function corresponding to the speech-recognized utterance phrase, provided that the analysis result of the analysis means is that the speech-recognized utterance phrase ends with a verbal stop or verb command form Means.

  According to such a configuration, it is analyzed whether or not a speech-recognized utterance phrase ends with a verbal stop or verb command form, and the result of this analysis is a speech-recognized utterance phrase A function corresponding to a speech-recognized utterance phrase is executed on the condition that it ends with a stop or verb command form.For example, in a conversation with the surroundings or a driver's monologue When a common utterance phrase ends with an adjective, the function corresponding to the speech-recognized utterance phrase is not executed, so that the wrong function execution due to conversation with the surroundings or the driver's monologue is reduced. Can do.

It is a figure which shows the structure of the speech recognition apparatus which concerns on one Embodiment of this invention. It is a flowchart of a control apparatus. It is a flowchart of a voice controller.

  The configuration of a speech recognition apparatus according to an embodiment of the present invention is shown in FIG. The voice recognition device is configured as a navigation device mounted on a vehicle. The navigation device includes a control device 10, a voice controller 20, an acoustic analysis unit 21, a voice recognition unit 22, and a grammar analysis unit 23. A detection signal is input to the control device 10 from a pressure sensor (not shown) that detects the pressure of the seating surface of each seat (driver's seat, front passenger seat, rear seat, etc.) of the vehicle. The acoustic analysis unit 21 is connected to a microphone 30 that collects sound in the passenger compartment.

  The control device 10 is configured as a computer including a CPU, a RAM, a ROM, a flash memory, an I / O, and the like. The CPU performs various processes according to a program stored in the ROM.

  Although not shown in FIG. 1, a map for inputting a position detector (vehicle speed sensor, gyroscope, GPS receiver) for specifying the current position of the vehicle, and map data to the control device 10. Data input device, audio equipment for music playback and radio reception, vehicle air conditioner for comfortable adjustment of vehicle interior temperature, communication device for communicating with external devices via wireless communication network, liquid crystal A display unit having a display such as a speaker, a speaker for outputting sound, and the like are connected.

  The control device 10 specifies the current position based on a signal for specifying the current position input from a position detector (vehicle speed sensor, gyroscope, GPS receiver) for specifying the current position of the vehicle. Specific processing, map display processing for displaying an image with the current position mark superimposed on a map around the vehicle on the display unit, audio device control processing for instructing audio device to play music or receive radio according to user operation, user A vehicle air conditioner control process for controlling the vehicle air conditioner according to the operation, a communication control process via the communication apparatus, and the like are performed.

  The voice controller 20 controls the acoustic analysis unit 21, the voice recognition unit 22, and the grammar analysis unit 23 in accordance with instructions from the control device 10. The audio controller 20 is configured as a computer including a CPU, RAM, ROM, flash memory, I / O, and the like, and the CPU performs various processes according to programs stored in the ROM.

  The acoustic analysis unit 21 continuously captures voice data from the microphone 30, sequentially analyzes the voice data to extract acoustic feature amounts, and obtains the acoustic feature amount time-series data obtained by the acoustic analysis, Dividing into several sections, the divided result is notified to the voice controller 20 as an analysis result.

  The voice recognition unit 22 acquires the analysis result of the acoustic analysis unit 21 via the voice controller 20, and performs processing for recognizing an utterance phrase spoken by the occupant based on the analysis result. That is, the speech recognition unit 22 recognizes a word sequence in the input speech by determining which word stored as dictionary data corresponds to each section divided by the acoustic analysis unit 21, and results of recognition. To the voice controller 20.

  The grammar analysis unit 23 acquires the recognition result of the voice recognition unit 22 via the voice controller 20, and performs grammar analysis processing of the utterance phrase uttered by the occupant based on the recognition result. That is, the grammar analysis unit 23 decomposes the word sequence in the input speech recognized by the speech recognition unit 22 into parts of speech, and analyzes whether the utterance phrase uttered by the occupant is a body stop or command form. Then, the analyzed result is notified to the voice controller 20.

  This navigation device always recognizes a speech phrase spoken by the user, and responds to the speech phrase recognized when the speech phrase recognized by the continuous speech recognition matches a predetermined registered phrase. Process to execute the function.

  Further, the navigation device according to the present embodiment analyzes whether or not the speech-recognized utterance phrase ends with a verbal stop or verb command form, and the result of this analysis is the speech-recognized utterance phrase. On the condition that the sentence ends with a verbal stop or verb command form, a process for executing a function corresponding to the speech-recognized utterance phrase is performed.

  Next, processing of the control device 10 and the voice controller 20 of the navigation device will be described. FIG. 2 shows a flowchart of the control device 10. FIG. 3 shows a flowchart of the voice controller 20.

  The navigation device is in an operational state when the ignition switch of the vehicle is turned on, the control device 10 performs the process shown in FIG. 2, and the voice controller 20 performs the process shown in FIG.

  First, the control apparatus 10 acquires vehicle information as shown in FIG. 2 (S100). In this embodiment, the detection signal from the pressure sensor provided in each seat (driver's seat, front passenger seat, rear seat, etc.) of the vehicle is acquired as vehicle information.

  Next, it is determined whether or not a passenger is present in the vehicle (S102). Specifically, it is determined whether a passenger is present in a seat other than the driver seat based on a detection signal of a pressure sensor provided in each seat of the vehicle (driver seat, passenger seat, rear seat, etc.).

  Here, if a passenger is in a seat other than the driver's seat of the vehicle, the determination in S102 is NO, and then the voice recognition start instruction is notified from the control device 10 to the voice controller 20 (S104). .

  On the other hand, as shown in FIG. 3, the voice controller 20 first determines whether or not a voice recognition start instruction has been received from the control device 10 (S200). When a voice recognition start instruction is not received from the control device 10, the determination in S200 is NO, and the determination in S200 is repeatedly performed.

  Then, when the voice controller 20 is notified of the voice recognition start instruction from the control device 10, the determination in S200 is YES, and the voice controller 20 instructs the acoustic analysis unit 21 to capture voice (S202).

  The acoustic analysis unit 21 continuously captures voice data from the microphone 30 in accordance with an instruction from the voice controller 20, and sequentially analyzes the voice data to extract an acoustic feature amount. The obtained acoustic feature amount time-series data is divided into several sections, and the divided result is notified to the voice controller 20 as an analysis result. The acoustic analysis unit 21 sequentially analyzes the audio data, and if the silent state continues for a certain period or longer, the acoustic analysis unit 21 considers that a series of utterance phrases of the occupant has been completed, and the analysis result is uttered in units of occupant utterance phrases. Notify the controller 20.

  Next, the voice controller 20 determines whether or not the analysis result from the acoustic analysis unit 21 has been received (S204). Here, when the analysis result from the acoustic analysis unit 21 is not received by the voice controller 20, the determination in S204 is NO, and the determination in S204 is repeatedly performed.

  When the analysis result from the acoustic analysis unit 21 is received by the voice controller 20, the determination in S204 is YES, and the voice controller 20 instructs the voice recognition unit 22 to start voice recognition (S206). Specifically, a command for instructing the start of voice recognition is transmitted to the voice recognition unit 22 together with the analysis result by the acoustic analysis unit 21.

  In response to an instruction from the voice controller 20, the voice recognition unit 22 performs a process of recognizing an utterance phrase uttered by an occupant based on an analysis result by the acoustic analysis unit 21. That is, the speech recognition unit 22 recognizes a word sequence in the input speech by determining which word stored as dictionary data corresponds to each section divided by the acoustic analysis unit 21, and results of recognition. To the voice controller 20 as a recognition result.

  Next, the voice controller 20 determines whether or not the recognition result from the voice recognition unit 22 has been received (S208). Here, when the recognition result from the voice recognition unit 22 is not received by the voice controller 20, the determination in S208 is NO, and the determination in S208 is repeatedly performed.

  When the recognition result from the voice recognition unit 22 is received by the voice controller 20, the determination in S208 is YES, and the voice controller 20 instructs the grammar analysis unit 23 to perform grammar analysis (S210). Specifically, a command for instructing grammar analysis is transmitted to the grammar analysis unit 23 together with the analysis result by the acoustic analysis unit 21.

  The grammar analysis unit 23 performs grammar analysis processing of the utterance phrase uttered by the occupant based on the recognition result by the voice recognition unit 22 in accordance with an instruction from the voice controller 20. That is, the grammar analysis unit 23 performs morphological analysis, specifies the part of speech of each word in the word sequence in the input speech recognized by the speech recognition unit 22, and the utterance when the utterance phrase uttered by the occupant ends with a noun The phrase is analyzed as a body stop, and when the utterance phrase uttered by the occupant ends with the verb command form, the utterance phrase is analyzed as a command form, and the analysis result is notified to the voice controller 20.

  For example, in the case of an utterance phrase “near convenience store”, it is decomposed into “near (adjective) _ (particle) _convenience store (noun)”. In this case, since the utterance phrase ends with “convenience store” which is a noun, the utterance phrase is analyzed to be a body stop.

  Also, in the case of the utterance phrase “music”, the utterance phrase cannot be further decomposed, and “music” is a noun, and therefore, the utterance phrase is analyzed as a body stop.

  In addition, in the case of the utterance phrase “raise the temperature of the air conditioner”, it is broken down into “air conditioner (noun) _no (particle) _temperature (noun) _raise (verb)”. Also in this case, since the utterance phrase ends with the verb command form “raise”, the utterance phrase is analyzed as the command form.

  Also, in the case of an utterance phrase “over FM”, it is broken down into “FM (noun) _over (verb)”. Also in this case, since the utterance phrase ends with “verb”, which is an instruction form of a verb, the utterance phrase is analyzed as an instruction form.

  Also, in the case of the utterance phrase “call Mr. A”, it is decomposed into “Mr. A (noun) _ (particle) _call me (verb)”. Also in this case, since the utterance phrase ends with “call” which is the verb command form, the utterance phrase is analyzed as the command form.

  However, in the case of the utterance phrase “previous car dangerous”, it is decomposed into “pre (noun) _no (particle) _car (noun) _dangerous (adjective)”. In this case, since the utterance phrase ends with the adjective “dangerous”, it is analyzed that the utterance phrase is neither a verbal stop nor a command form.

  Also, in the case of the utterance phrase “neighboring sound noisy”, it is broken down into “neighboring (noun) _no (particle) _sound (noun) _noisy (adjective)”. Also in this case, since the utterance phrase ends with the adjective “noisy”, it is analyzed that the utterance phrase is neither a verbal stop nor a command form.

  In the case of an utterance phrase “rain is heavy”, “rain (noun) _ is (particle) _heavy (adjective)”. Also in this case, since the utterance phrase ends with the adjective “violent”, it is analyzed that the utterance phrase is neither a verbal stop nor a command form.

  Next, the voice controller 20 determines whether or not the analysis result from the grammar analysis unit 23 has been received (S212). Here, when the analysis result from the grammar analysis unit 23 is not received by the voice controller 20, the determination in S208 is NO and the determination in S212 is repeatedly performed.

  When the analysis result from the grammar analysis unit 23 is received by the voice controller 20, the determination in S212 is YES, and the voice controller 20 determines whether the utterance phrase is a body stop or command based on the analysis result from the grammar analysis unit 23. It is determined whether or not it is a shape (S214).

  Here, when it is determined based on the analysis result from the grammar analysis unit 23 that the utterance phrase is body stop or command form, the determination in S214 is YES, and the voice controller 20 adopts the recognition result to the control device 10. A notification to this effect is sent (S216). Specifically, a command instructing the adoption of the recognition result is transmitted to the control device 10 together with the result recognized by the voice recognition unit 22.

  If it is determined that the utterance phrase is neither body speech nor command based on the analysis result from the grammar analysis unit 23, the determination in S214 is NO and the voice controller 20 rejects the recognition result to the control device 10. A notification to this effect is sent (S218). Specifically, a command for instructing not to adopt the recognition result of the voice recognition unit 22 is transmitted to the control device 10.

  Further, as shown in FIG. 2, the control device 10 notifies the voice controller 20 of a voice recognition start instruction in S <b> 104 and then receives a notification from the voice controller 20 to adopt the recognition result. Is determined (S106).

  Here, when the voice controller 20 receives a command instructing the adoption of the recognition result transmitted from the voice controller 20, the determination in S106 is YES, and the voice controller 20 responds to the recognition result of the voice recognition unit 22. The function is executed (S108).

  In other words, the grammar analysis unit 23 stops speaking the utterance phrase like utterance phrases such as “Nearby convenience store”, “Music”, “Raise the temperature of the air conditioner”, “Call FM” and “Call Mr. A”. Alternatively, with respect to what is analyzed as being in the command form, a function corresponding to the speech phrase that has been voice-recognized is executed.

  When the voice controller 20 receives a command that instructs not to adopt the recognition result transmitted from the voice controller 20, the determination in S106 is NO and the voice controller 20 recognizes the recognition result of the voice recognition unit 22. The process returns to S106 without executing the function corresponding to.

  In other words, for utterance phrases such as “previous car dangerous”, “adjacent sound annoyance”, and “rainy”, the grammar analysis unit 23 analyzes that the utterance phrase is neither body stop nor command form. The function corresponding to the utterance phrase recognized by voice is not executed.

  If the passenger is not in a seat other than the driver's seat of the vehicle, the determination in S102 is YES, and the control device 10 performs this processing without transmitting a voice recognition start instruction to the voice controller 20. Exit. Therefore, the function corresponding to the speech-recognized utterance phrase is not executed.

  According to the above configuration, the navigation device analyzes whether or not the speech-recognized utterance phrase ends with a verbal stop or verb command form, and the result of this analysis is the speech-recognized utterance. On the condition that the completed phrase ends with a verbal stop or verb command form, for example, conversation with the surroundings or driver If an utterance phrase that often appears in a single word of a word ends with an adjective, the function according to the spoken phrase that was recognized by the voice is not executed. Can be reduced.

  In addition, when analyzing whether a speech-recognized utterance phrase ends with a verbal stop or verb command form, this navigation device divides the speech-recognized utterance phrase into parts of speech and is recognized as speech. When the last part of speech of a spoken phrase is a noun or verb command form, it can be analyzed that the speech recognized speech phrase ends with a verbal stop or verb command form.

  Further, the navigation device acquires a detection signal from a pressure sensor that detects the pressure of the seating surface of each seat of the vehicle as vehicle information for determining whether or not a passenger is present in the vehicle. It is determined whether or not a passenger is present in the vehicle based on the above, and when it is determined that there is no passenger in the vehicle, the function according to the speech-recognized utterance phrase is not executed. It is possible to further reduce erroneous function execution due to conversation with surroundings.

  The present invention is not limited to the above-described embodiment, and various modifications can be made as follows without departing from the spirit of the present invention.

  For example, in the above embodiment, it is determined whether there is a passenger in a seat other than the driver's seat based on a detection signal from a pressure sensor provided in each seat of the vehicle (driver's seat, front passenger seat, rear seat, etc.). However, for example, it may be determined whether there is a passenger in a seat other than the driver's seat based on a signal detected by a door opening sensor that detects the open / closed state of each door of the vehicle. Specifically, when the door opening sensor detects that the door of the driver's seat and the door other than the driver's seat have been opened and closed between the time when the door of the vehicle is unlocked and the time when the vehicle starts to travel. It can be determined that there is a passenger in a seat other than the driver's seat, and that the door of the driver's seat has been opened and closed between the time the vehicle door is unlocked and the time the vehicle starts to run. If it is detected by the door opening sensor that the door other than the driver's seat is not opened / closed even though it is detected by the sensor, it can be determined that there is no passenger in the seat other than the driver's seat.

  Moreover, in the said embodiment, although this voice recognition apparatus was comprised as a navigation apparatus mounted in a vehicle, it is not limited to what is mounted in a vehicle, For example, a smart phone, a mobile telephone, a portable information terminal, etc. Can also be used. When configured using such a portable terminal, vehicle information for determining whether or not a passenger is present in the vehicle is wirelessly transmitted from the vehicle side, and the portable terminal receives the vehicle information. It is preferable to obtain it.

  The correspondence relationship between the configuration of the above embodiment and the configuration of the claims will be described. The acoustic analysis unit 21, the speech recognition unit 22, and the grammar analysis unit 23 correspond to analysis means, and S214, S216, S218, and S106. And S108 correspond to function execution means, S100 corresponds to vehicle information acquisition means, and S102 corresponds to passenger determination means.

DESCRIPTION OF SYMBOLS 10 Control apparatus 20 Voice controller 21 Acoustic analysis part 22 Speech recognition part 23 Grammar analysis part 30 Microphone

Claims (4)

Speech that recognizes a speech phrase spoken by the user and performs a function corresponding to the speech phrase that has been speech-recognized when the speech phrase recognized by speech recognition matches a predetermined registered phrase A recognition device,
Analyzing means for analyzing whether the speech-recognized utterance phrase ends with a verbal stop or verb command form;
The analysis result of the analyzing means executes a function corresponding to the speech-recognized utterance phrase on the condition that the speech-recognized uttered phrase ends with a verbal stop or verb command form. A speech recognition apparatus comprising: a function executing unit;   The analysis means divides the speech-recognized utterance phrase into parts of speech, and when the last part-of-speech of the speech-recognized utterance phrase is a noun or verb command form, The speech recognition apparatus according to claim 1, wherein the speech recognition apparatus analyzes that the command ends with a verb command form. Vehicle information acquisition means for acquiring vehicle information for determining whether or not a passenger is present in the vehicle;
Passenger determination means for determining whether a passenger exists in the vehicle based on the vehicle information acquired by the vehicle information acquisition means,
3. The function execution unit does not execute a function corresponding to the speech-recognized utterance phrase when it is determined by the passenger determination unit that no passenger exists in the vehicle. The speech recognition apparatus described in 1.   The vehicle information for determining whether or not a passenger is present in the vehicle is a detection signal from a pressure sensor that detects a pressure of a seating surface of each seat of the vehicle. Voice recognition device.

Images