JP3963698B2 - Spoken dialogue system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a voice dialogue system that automatically performs control of vehicle electrical components such as an air conditioner, a power window, and a car navigation system, an audio device, and a PDA, based on a dialogue with a user.
[0002]
[Prior art]
In devices with complicated operations such as car navigation systems and audio devices, the operation of the devices is simplified by using a voice interaction system. In this spoken dialogue system, the system utters a question created based on a predetermined step, recognizes a user's answer to the question, and creates a new utterance based on the recognition result. Information necessary for the operation is obtained from the user.
[0003]
[Problems to be solved by the invention]
However, in such a conventional voice interaction system, it is necessary to take a voice recognition timing on the user side in order to activate the voice recognition device. Actually, the conventional apparatus is configured to activate the voice recognition apparatus by pressing or continuing to press a switch or button indicating “recognition start” or “speaking” in advance. Such a configuration not only makes the operation complicated, but also imposes a burden on the user other than the driving operation, and there is a risk of inducing a driving error.
[0004]
The present invention has been made for the purpose of solving the above-described drawbacks of the conventional apparatus, and the device automatically recognizes the start of the dialog system, rather than the user side acting on the dialog system to start the recognition. It is an object of the present invention to provide a spoken dialogue system that can be used.
[0005]
[Means for Solving the Problems]
This onset Ming, in order to solve the above problems, for controlling the navigation system by voice dialogue, a voice dialogue system in the vehicle, a sensor for detecting that the parking brake is turned on, the sensor An audio output device that outputs a word prompting registration of billing information for the parking lot when receiving a signal indicating that the vehicle has entered the parking lot from the navigation system. Provide an interactive system.
[0006]
The sensor may be a sensor which detects that the path parking brake is turned on.
[0007]
When output signals from these sensors are input to the system, the speech recognition device is automatically activated, and words determined in advance corresponding to the input signals are output from the speech output device. This is a word that gives the user a voice dialogue from the system side. For example, if the sensor detects the engine on state, it will say "What do you do?" is there. Therefore, when the user answers, for example, “I want to turn on the air conditioner” to this word, the voice recognition device recognizes this utterance, and the voice output device can start a predetermined dialogue routine based on this recognition. .
[0008]
In addition, when the sensor detects that the engine is on and further detects that the driver has boarded, the voice recognition device is activated, and the voice output device triggers a dialogue such as "Where are you going?" Output the word. When the user answers this question, for example, “I want to go to ***”, the voice recognition device recognizes this, and the voice output device can start a predetermined dialogue routine based on this recognition.
[0009]
As described above, in the present invention, the speech recognition apparatus is activated by the output signals from various sensors provided in the vehicle, and words that trigger the dialogue are automatically output from the dialogue system side. Therefore, the user does not have to start the speech recognition apparatus by operating the start button or the like. Therefore, the voice conversation can be started smoothly.
[0010]
In addition, by adding words that do not make any special meaning such as “Mr. XX” or “Moshimoshi” to the head of the words output from the voice output device, the user can more easily hear questions from the system.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing the structure of a spoken dialogue system according to an embodiment of the present invention. In the figure, 1 is a voice recognition device, 2 is a voice output device, 3 is a sensor provided in each part of the vehicle, and 4 is a vehicle electrical component to be controlled by this interactive system. The speech recognition device 1 is a device for recognizing a user's utterance as a word, and generally includes a speech recognition engine and a recognition dictionary. The voice output device 2 includes a voice recognition engine, a dialogue database, and a voice synthesis engine, and asks questions to the user according to a program determined in advance based on the user's utterance recognized as words by the voice recognition device 1. A predetermined interactive routine is executed by creating and outputting this as synthesized speech.
[0012]
The information acquired by the voice recognition device 1 as described above is sent to the vehicle electrical component 4 to operate the device in accordance with the user's intention. The vehicle electrical component 4 is, for example, an air conditioner, a power window, a navigation system, an audio / video device, an information terminal device, or the like.
[0013]
In the present apparatus, signals indicating various states of the vehicle are output to the voice recognition device 1 and the voice output device 2 from the sensors 3 provided in each part of the vehicle. This sensor 3 is, for example, a sensor that detects that the engine is turned on, a sensor that detects that the engine is turned on and the driver gets on, a sensor that detects that the fuel has become a certain amount or less, VICS, A sensor for detecting traffic jam information, a sensor for detecting incoming mail, a sensor for detecting that the parking brake is turned on, and the like.
[0014]
Output signals from the sensor 3 are simultaneously input to the audio output device 2. By this input, the voice output device 2 outputs a predetermined word corresponding to the input signal as a synthesized voice.
[0015]
FIG. 2 shows the relationship between the type of information output from the sensor 3 and a predetermined audio output. In FIG. 2A, when a signal indicating that the engine is turned on is output from the sensor 3, a question “What are you doing?” Is output from the audio output device 2. Show. On the other hand, when the user answers, for example, “I want to turn on the air conditioner”, the voice recognition device 1 recognizes this utterance and starts a dialogue routine for controlling the air conditioner in the voice output device 2.
[0016]
If the user utters words such as “not necessary”, the speech recognition apparatus 1 recognizes this and automatically ends the dialogue. Alternatively, when the voice recognition device 1 detects that there is no voice input for a certain period of time, the dialogue may be automatically terminated. Further, the dialogue may be automatically ended by uttering words such as “No need” or “End guidance” in the voice output device 2.
[0017]
In the example shown in FIG. 2B, an example in which a car navigation system is incorporated as an in-vehicle device, and information indicating that the engine is turned on and the driver gets on is output from the sensor 3. Show. In this case, since the probability that the driver moves to somewhere is very high, the voice output device 2 asks "Where should I guide you?" When the user answers “I want to go to ***” or the like, the speech recognition apparatus 1 recognizes this and automatically starts a dialogue for setting the destination of the navigation system.
[0018]
If there is no need for voice guidance, the dialogue is automatically terminated when the user's voice is not input for a certain period of time or when the user says “not necessary”, as in the case of (a). To do.
[0019]
As a result, the user can set the destination in a routine manner when getting on, and the user does not need to memorize the destination setting command.
[0020]
In the example shown in (c) of FIG. 2, a car navigation system is incorporated as an in-vehicle device as in the case of (b), and the remaining amount of gasoline is detected by the sensor 3. In this case, when the remaining amount of gasoline is below a certain value, the voice output device 2 will ask the question “While gasoline is low. Start the conversation. This eliminates the need for the user to drive while caring for the remaining amount of fuel.
[0021]
The example shown in (d) of FIG. 2 is a modification of (c), in which the remaining amount of gasoline is detected by the sensor 3 and the distance to the destination plus several tens of times in consideration of the distance to the destination and fuel consumption. When it is within km, a warning such as "Can't reach the destination with the current fuel. Be sure to refuel at the next stand." Start a dialogue for navigation by asking questions.
[0022]
This eliminates the need for the user to drive while paying attention to the fuel and the fuel replenishment destination even in long-distance driving.
[0023]
In the example shown in FIG. 2 (e), the traffic information from the VICS is used as a trigger for the dialogue in a vehicle equipped with a navigation system capable of obtaining VICS (road traffic information communication system) information. In other words, when traffic information is input from VICS, the system asks the question “This traffic jam has already occurred. Do you want to detour?” Before reaching the road where the traffic has occurred. Start a dialogue for navigation.
[0024]
This eliminates the need for the user to drive while confirming the traffic jam, so that he can concentrate on driving.
[0025]
In the example shown in FIG. 2 (f), for example, in a vehicle equipped with an information terminal having a function of detecting the reception of an email, when the email is received, the system side says, “Mail has arrived. ”To start a dialogue system for reading a mail.
[0026]
FIG. 3 is a diagram showing a flow in a case where an incoming e-mail is triggered by a dialogue and the dialogue is advanced. First, when the sensor receives the incoming mail information, the dialogue system asks the question “A mail has arrived. Do you want to read it out?” To form the start timing of the mail reading conversation (steps) S10).
[0027]
In this case, for example, an e-mail reading function as shown in the figure is executed according to the user's reaction “yes”, “after”, “title”, “from whom?”. For example, if the user replies “yes”, the user enters an interactive routine for mail processing, and reads the contents of the mail by the synthesized voice (step S11). If the user replies “after”, the voice conversation is once ended on the system side (step S12). If the user answers “What is the title?”, The title of the mail is read out by the synthesized voice (step S13). If the user replies “From whom?”, The name of the sender is read out by the synthesized voice (step S14). It is preferable that a nickname or the like can be registered in advance for the sender of the mail.
[0028]
In step S13, when the system reads out the title of the mail, if the user further answers “read”, the system reads out the content of the mail by the synthesized voice (step S15). When the user's response to step S13 is “next?”, The next title is read out by the synthesized voice in step S16. If the user's response to step S13 is "I'll do it later", the dialog system is once terminated in step S17.
[0029]
In step S14, the sender of the mail is read out, and when the user responds “read”, the system reads out the contents of the mail in synthesized voice in step S18. If the user response is "What is next?", In step S19, the system reads the name of the next sender in a synthesized voice. If the user's response is “I'll do it later”, the system once ends the voice conversation in step S20.
[0030]
As described above, the user can not only automatically know the incoming mail, but can select the mail to be read out by the title or sender, which is very convenient.
[0031]
FIG. 4 is a diagram showing still another embodiment of the present invention, and shows a flow of voice interaction for a vehicle having a function capable of registering parking lot billing information and a navigation function. As a function for registering the billing information, the billing information can be registered in advance, but in the illustrated embodiment, when the vehicle is received and the parking brake is set, an inquiry is made to prompt the user to register the billing information. From the system side.
[0032]
First, when the navigation system recognizes that the vehicle has entered the parking lot in step S30, the time at that time, that is, the entry time is stored in the system. Next, when the parking brake is turned on in step S31, the system asks "Do you want to register the parking fee system?" And asks the user for a dialog for charging information registration guidance. Prompt.
[0033]
When the parking fee system is registered according to the guidance of the voice dialogue system, the engine is turned off (step S32), and then the engine is turned on again (step S33), the system determines that this state is a delivery, and Output parking time and toll information calculated based on time. This output is, for example, a guidance by synthetic voice such as “Parking time is *** minutes, so it will be XX circle”.
[0034]
As a result, the user can know the parking fee in advance before leaving, which is very convenient.
[0035]
FIG. 5 is a diagram showing still another embodiment of the present invention, in which the voice dialogue system of the present invention is applied to a navigation system having toll road toll information or a navigation system capable of registering toll road toll information. The embodiment in the case where it did is shown.
[0036]
In this system, when the traveling vehicle passes the entrance of the toll road, the entrance information is stored as shown in step S40, and when the toll gate approaches as shown in step S41, the system side “ “*** yen” will be output, and this will be used as the voice recognition start timing. As a result, the user can ask the system questions such as “What is the next toll booth” and “How much is next?” Without pressing the voice recognition start button.
[0037]
When the traveling speed of the vehicle becomes 0, the system recognizes that the fee has been paid (step S42), and stops voice recognition. When the vehicle resumes running, the system recognizes that the fee payment has ended (step S43), and starts voice recognition.
[0038]
As a result, the user can know the toll in advance before the toll gate on the toll road, so that preparation for payment can be made smoothly and the time for passing through the toll gate can be shortened.
[0039]
While various embodiments of the present invention have been described above with reference to FIGS. 1 to 5, various modifications can be made to the above embodiments. For example, in a spoken dialogue system having all the above functions, the priority when multiple conditions overlap, the necessity, and the necessity may be set arbitrarily. For example, when incoming mail and fee guidance occur at the same time, priority is given to fee guidance, and when a series of conversations are completed, email guidance is set, or only email without fee guidance is performed. It is also possible to set to read.
[0040]
Furthermore, in the voice dialogue system having all the above functions, it is possible to select a function for performing guidance and asking questions. For example, it is possible to select whether or not fee guidance is necessary and whether or not fuel warning is necessary.
[0041]
Further, as a drawback of the voice dialogue system, there is a high probability that the user misses the first part of the guidance from the system side. Therefore, as a trigger for the dialogue, words that are meaningless to the content of the dialogue, such as “Mr. XXX”, “Moshimoshi”, “I am a little rude”, may be issued at the beginning of the dialogue. The meaningless words that are emitted may be the same word each time, or a plurality of variations may be provided.
[0042]
Furthermore, it is possible to change the language that triggers the dialogue according to the information to be guided. For example, when the traffic situation changes due to VICS traffic information or the like, it is possible to add a word such as “Breaking news” to the beginning of the sentence.
[0043]
Furthermore, the text that triggers the conversation may be changed according to the urgency of the information. For example, in toll road fee guidance, when traveling slowly, the travel section and fee guidance may be provided in detail, and when traveling at high speed, only the fee guidance may be required. As an example, when driving in a traffic jam or at low speed, the system says, “Soon it will be a toll booth. It ’s a circle. ”
[0044]
Furthermore, immediately after the above meaningless call, if the user responds with a `` wait a moment '', `` wait '', etc., this is recognized on the system side, the system is shifted to the standby state while maintaining the recognition state, It is also possible to suspend the dialogue. In this case, the user's next utterance, “yes”, “what?” Is recognized and the dialogue is continued. Alternatively, after a certain period of time has passed with the dialogue pending, it is also effective to ask again the words such as “Are you sure you want to guide me” or “Is it still?” It is also effective to indicate what information is to be provided after a certain period of time, for example, “Guidance ends. Furthermore, when it is determined that the information is no longer necessary while the dialogue is on hold, it is possible to cancel the on-hold state of the dialogue. For example, if you leave the parking lot with the parking fee guide on hold, the fee guide is not required, so you will be notified of the location by giving a notice such as "Release the reserved fee guide". You may make it cancel.
[0045]
【The invention's effect】
As described above, as shown in various embodiments, in the voice dialogue system according to the present invention, a voice trigger is used to output a dialogue trigger word according to an output signal from a sensor provided in each part of the vehicle. I can do it. Therefore, since the user does not need to create a dialog for the voice dialogue system from himself, he can concentrate on driving. Also, the necessary voice guidance timing is not missed.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a voice interaction system according to an embodiment of the present invention.
FIG. 2 is a diagram for explaining the operation of the system shown in FIG. 1;
FIG. 3 is a diagram showing a flow when a mail reading function is executed by the system according to the embodiment of the present invention.
FIG. 4 is a diagram showing a flow in the case of executing a parking fee guide function by the system according to the embodiment of the present invention.
FIG. 5 is a diagram showing a flow in a case where toll road fee guidance is executed by the system of one embodiment of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Voice dialogue system 2 ... In-vehicle apparatus 11 ... Voice recognition apparatus 12 ... Voice output device 13 ... Sensor of each part of vehicle

Claims (1)

A voice dialogue system in a vehicle for controlling a navigation system by voice dialogue,
A sensor that detects that the parking brake is on;
A voice output device that receives an output signal from the sensor and receives a signal indicating that a vehicle has entered the parking lot from the navigation system;
A spoken dialogue system in a vehicle comprising:

Images