Notification device, notification method, notification program, and recording medium

Abstract

Aside from words used to control specific devices, this system makes it possible to make speakers aware of inappropriate verbal tics. [Solution] The notification device acquires non-control word information indicating words not used to control a predetermined device based on the sound input means, stores the non-control word information in a memory unit, identifies words that are used more times than a predetermined number of times based on the non-control word information, and if the identified word corresponds to a word predetermined as an inappropriate word, it notifies notification information indicating the identified word.

Description

Technical Field

[0001] This application relates to a method for providing notification when a specific word is spoken.

Background Art

[0002] For example, Patent Document 1 discloses a notification device that provides notification when a user uses voice data that the user desires to be notified of. Specifically, comparative voice data such as "···te yu~ka~" and "oops" is input and stored in this notification device. When the voice signal of the user input by the microphone later matches the comparative voice data, the notification device provides notification.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] There are devices that recognize commands from a user by voice recognition and operate based on those commands. It is not reasonable to provide notification just because the word uttered as a command to such a device matches a specific word. Also, if notification is provided just because the uttered word happens to match a specific word, there is a possibility that excessive notification will occur. On the other hand, an inappropriate word may become a habit without being noticed. It is expected that by pointing out such a habit, it will provide an opportunity to correct the inappropriate habit.

[0005] The present invention has been made in view of the above points, and an example of its problem is to provide a notification device, a notification method, a notification program, and a recording medium that enable a speaker to notice an inappropriate habit when there is such a habit, except for words for controlling a specific device.

Means for Solving the Problems

[0006] To solve the above problems, the invention described in claim 1 is a notification device comprising: an acquisition means for acquiring non-control word information indicating words not used for controlling a predetermined device based on sound input using a sound input means; a storage control means for storing the acquired non-control word information in a storage means; an identification means for identifying a word that is used more than a predetermined number of times based on the stored non-control word information; and a notification control means for notifying notification information indicating the identified word when the identified word corresponds to a word predetermined as an inappropriate word, wherein the inappropriate word includes at least one of rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms; the storage means stores inappropriate word information indicating inappropriate words according to the situation in which the speaker is placed; and the notification control means, based on the inappropriate word information indicating an inappropriate word corresponding to the identified word, (a) causes the notification information to be notified by an output means connected to the predetermined device, or (b) causes the notification information to be transmitted to a terminal device different from the predetermined device and used by the speaker.

[0007] The invention described in claim 2 is a notification device comprising: an acquisition means for acquiring non-control word information indicating words not used for controlling a predetermined device based on sound input using a sound input means; a storage control means for storing the acquired non-control word information in a storage means; an identification means for identifying a word that is used more than a predetermined number of times based on the stored non-control word information; and a notification control means for notifying notification information indicating the identified word when the identified word corresponds to a word predetermined as an inappropriate word, wherein the inappropriate word includes at least one of rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms; the storage means stores inappropriate word information indicating inappropriate words according to the situation in which the speaker is placed; and the notification control means notifies notification information according to the situation based on the inappropriate word information indicating an inappropriate word corresponding to the identified word.

[0008] The invention described in claim 4 is a notification method performed by a computer, comprising: an acquisition step of acquiring non-control word information indicating words not used to control a predetermined device based on sound input using sound input means; a storage control step of storing the acquired non-control word information in a storage means; an identification step of identifying words that are used more than a predetermined number of times based on the stored non-control word information; and a notification control step of broadcasting notification information indicating the identified words if the identified words correspond to words predetermined as inappropriate words, wherein the inappropriate words The inappropriate words include at least one of the following: rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms. The storage means stores inappropriate word information indicating words that are inappropriate depending on the situation in which the speaker is placed. The notification control step is a notification method characterized by (a) causing the notification information to be broadcast by an output means connected to the predetermined device, or (b) causing the notification information to be transmitted to a terminal device different from the predetermined device and used by the speaker, based on the inappropriate word information indicating words that correspond to the identified words.

[0009] The invention described in claim 5 is a notification method performed by a computer, comprising: an acquisition step of acquiring non-control word information indicating words not used to control a predetermined device based on sound input using sound input means; a storage control step of storing the acquired non-control word information in a storage means; an identification step of identifying a word that is used more than a predetermined number of times based on the stored non-control word information; and a notification control step of notifying notification information indicating the identified word if the identified word corresponds to a word predetermined as an inappropriate word, wherein the inappropriate word includes at least one of rude words, offensive words, dirty words, words that give a negative impression, slang and discriminatory terms; the storage means stores inappropriate word information indicating inappropriate words according to the situation in which the speaker is placed; and the notification control step is characterized in that it notifies notification information according to the situation based on the inappropriate word information indicating an inappropriate word corresponding to the identified word.

[0010] The invention described in claim 6 is a notification program characterized in that a computer functions as: an acquisition means for acquiring non-control word information indicating words not used to control a predetermined device based on voice input using a sound input means; a storage control means for storing the acquired non-control word information in a storage means; an identification means for identifying words that are used more than a predetermined number of times based on the stored non-control word information; and a notification control means for notifying notification information indicating the identified word if the identified word corresponds to a word predetermined as an inappropriate word, wherein the inappropriate word includes at least one of rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms; the storage means stores inappropriate word information indicating inappropriate words according to the situation in which the speaker is placed; and the notification control means, based on the inappropriate word information indicating an inappropriate word corresponding to the identified word, (a) causes the notification information to be notified by an output means connected to the predetermined device, or (b) causes the notification information to be transmitted to a terminal device different from the predetermined device and used by the speaker.

[0011] The invention described in claim 7 is a notification program characterized in that a computer functions as an acquisition means for acquiring non-control word information indicating words not used to control a predetermined device based on voice input using a sound input means; a storage control means for storing the acquired non-control word information in a storage means; a identification means for identifying words that are used more than a predetermined number of times based on the stored non-control word information; and a notification control means for notifying notification information indicating the identified word when the identified word corresponds to a word predetermined as an inappropriate word, wherein the inappropriate word includes at least one of rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms, the storage means stores inappropriate word information indicating inappropriate words according to the situation in which the speaker is placed, and the notification control means notifies the notification information according to the situation based on the inappropriate word information indicating an inappropriate word corresponding to the identified word.

[0012] The invention described in claim 8 is a recording medium characterized by having the notification program recorded in a computer-readable format. [Brief explanation of the drawing]

[0013] [Figure 1] This is a block diagram showing an example of the schematic configuration of a notification device according to an embodiment. [Figure 2] (a) is a diagram showing an example of a vehicle according to the embodiment. (b) is a block diagram showing an example of a navigation device according to the embodiment. [Figure 3] (a) is a diagram showing an example of information stored in the inappropriate word database according to the embodiment. (b) is a diagram showing an example of information stored in the speech history database according to the embodiment. [Figure 4] This figure shows an example of notification provided by a navigation device according to an embodiment. [Figure 5] This is a flowchart showing an example of processing by the navigation device according to the embodiment.

Best Mode for Carrying Out the Invention

[0014] First, a mode for carrying out the present application will be described with reference to FIG. 1. FIG. 1 is a block diagram showing an example of the schematic configuration of an information device according to an embodiment.

[0015] As shown in FIG. 1, an information device D according to an embodiment includes an acquisition unit 1, a storage control unit 2, a specification unit 3, and an information notification control unit 4.

[0016] In this configuration, the acquisition unit 1 acquires control word information indicating words used for controlling a predetermined device and non-control word information indicating words not used for controlling the predetermined device based on the voice input by using the voice input unit 5. The predetermined device may be the same as or different from the information device D.

[0017] The storage control unit 2 stores the non-control word information acquired by the acquisition unit 1 in the storage unit 6.

[0018] The specification unit 3 specifies words whose usage frequency is higher than a predetermined number based on the non-control word information stored in the storage unit 6.

[0019] When the word specified by the specification unit 3 corresponds to a word predetermined as an inappropriate word, the information notification control unit 4 causes the information notification device to notify information indicating the specified word.

[0020] As described above, among the words not used for controlling the predetermined device, if an inappropriate word is spoken more than a predetermined number of times, that word is notified. Therefore, except for the words for controlling a specific device, it is possible to make the speaker aware of an inappropriate habit of speaking.

Examples

[0021] Next, specific examples corresponding to the above-described embodiments will be described with reference to FIGS. 2 to 5. The examples described below are examples when the embodiments are applied to a navigation device that can be mounted on a vehicle.

[0022] [1. Configuration of Navigation Device] FIG. 2(a) is a diagram showing an example of a vehicle according to an embodiment. FIG. 2(b) is a block diagram showing an example of a navigation device according to an embodiment. FIG. 3(a) is a diagram showing an example of information stored in an inappropriate word database according to an embodiment. FIG. 3(b) is a diagram showing an example of information stored in a conversation history database according to an embodiment. FIG. 4 is a diagram showing an example of notification by the navigation device according to the embodiment. FIG. 5 is a flowchart showing an example of processing by the navigation device according to the embodiment.

[0023] As shown in FIG. 2, the navigation device 10 according to the embodiment can be mounted on the vehicle 30. The navigation device 10 may mainly perform processing for guiding the vehicle 30 to a destination.

[0024] As shown in FIG. 2(b), the navigation device 10 includes a control unit 11, a storage unit 12, a communication unit 13, a display 14, a speaker 15, a microphone 16, an input unit 17, and an interface unit 18. The control unit 11 to the interface unit 18 are connected via a bus 19.

[0025] The control unit 11 includes a CPU, a ROM, a RAM, etc. By the CPU reading out and executing various programs stored in the ROM and the storage unit 12, the control unit 11 executes various processes described later.

[0026] The storage unit 12 is composed of non-volatile memory such as a hard disk or flash memory. The storage unit 12 stores various programs for controlling the navigation device 10. These programs may be read from a recording medium via a drive device (not shown), or downloaded from a predetermined server device via a network.

[0027] Furthermore, road data is stored in the memory unit 12. The road data is data related to roads. For example, the road data may include the type, distance, and shape of each road, as well as the connection status between roads and the latitude and longitude of the road connection points.

[0028] Furthermore, the memory unit 12 stores command words that represent words for controlling the operation of the navigation device 10. When a passenger in the vehicle 30 speaks one command word or a combination of command words, the navigation device 10 recognizes the command from the passenger through voice recognition and operates according to that command. Examples of operations of the controlled navigation device 10 include searching for a destination, displaying a map, and providing route guidance. Command words may be expressed as text information, for example. For example, the memory unit 12 may store a voice command database, a genre database, a location information database, etc. These databases contain command words. The voice command database stores command words that represent words as commands to the navigation device 10, and information that indicates the operation of the navigation device 10 corresponding to that command, in association with each other. Examples of command words include "name," "address," "surroundings," "current location," "view destination map," "go here," "stop by," "view map," and "clear route." The genre database stores genre information that represents each genre of facility as command words. Examples of genres include "convenience store," "gas station," "parking lot," and "family restaurant." The location information database stores, as command words, name information indicating the name of each location, information indicating the genre of that location, and information indicating the longitude and latitude of that location, all associated with each other. The navigation device 10 may also be capable of playing music, playing videos, receiving television broadcasts, receiving radio broadcasts, etc. A database containing command words for controlling these processes may be stored in the storage unit 12.

[0029] Furthermore, the memory unit 12 stores an inappropriate word database, a speaker database, and a speech history database. The inappropriate word database stores an inappropriate word dictionary containing one or more inappropriate words that indicate inappropriate language. The inappropriate word dictionary may be represented, for example, by textual information. Inappropriate words may include at least one of the following: a word that is rude to the person to whom it is directed, a word that offends the person to whom it is directed, a dirty word, a word that gives a negative impression, slang, a discriminatory term, etc. The inappropriate word database may store multiple inappropriate word dictionaries that indicate inappropriate language according to multiple different situations in which the speaker is placed. Examples of situations in which the speaker is placed include when the speaker is speaking as the driver of vehicle 30, when the speaker is speaking as a passenger in vehicle 30 rather than the driver, and when a parent (or simply an adult) is speaking to a child. As shown in Figure 3(a), the inappropriate word database may store, for example, separate inappropriate word dictionaries for the driver, passenger, and parent. Words that are inappropriate for a driver to use may include, for example, vulgar language expressing complaints or dissatisfaction with other vehicles around vehicle 30, their drivers or pedestrians, or traffic conditions. Examples of such words include "What the hell," "You bastard," "You're in the way," "Tch" (clicking tongue), and sighs. Words that are inappropriate for a passenger to use may include, for example, words expressing complaints or dissatisfaction with the driver of vehicle 30's driving, or words that burden the driver. Examples of such words include "I knew there would be a traffic jam," "Aren't we there yet?", and "Are we going to make it?". Words that are inappropriate for a parent to use may include, for example, slang that the parent doesn't want their child to learn, or words used to scold a child harshly. Examples of such words include "That's bad," "Really," "Quite," "No," "Hurry up!", "Wait a minute," and "Do it." The inappropriate word database may store a standard inappropriate word dictionary that indicates words that are inappropriate in any situation. Among the inappropriate word dictionaries stored in the inappropriate word database, the inappropriate word dictionary used for reporting inappropriate verbal habits may be pre-selectable by the passengers of vehicle 30.For example, it may be possible to select one or more dictionaries of inappropriate words.

[0030] The inappropriate word database may store at least one of a notification information template and a notification mode associated with each inappropriate word dictionary. The notification information template is a template for notification information that is issued when an inappropriate word spoken by a passenger of vehicle 30 is a catchphrase of that passenger. This template may be, for example, text information or an audio feature vector of the notification information. One notification information template may be stored for each inappropriate word dictionary, or one notification information template may be stored for each of the one or more similar inappropriate words included in the inappropriate word dictionary. In the case of an inappropriate word dictionary for drivers, the template may be, for example, "Now, now, calm down a little. XXX is not good." Here, "XXX" is an inappropriate catchphrase. This template message is thought to have the effect of calming the driver. In the case of an inappropriate word dictionary for passengers, the template may be, for example, "XXX might be a little harsh to say to a driver." In a dictionary of inappropriate words for parents, for words that strongly scold children, the template might be, for example, "XXX is a somewhat overbearing way of speaking to a child." The notification mode indicates by what means the message will be notified. For example, there may be "direct" and "speaker terminal" as notification modes. "Direct" indicates that notification will be made directly by the navigation device 10. "Speaker terminal" indicates that notification will be made by a terminal device carried by the speaker. Examples of portable terminal devices include smartphones, tablet computers and other personal information terminals, mobile phones, PDAs (Personal Digital Assistants), etc. For example, notification information may be sent to the terminal device by short message, email, etc.

[0031] The speaker database stores information about speakers. For example, the speaker database may store a speech feature vector for each speaker, associated with a speaker ID. The speaker ID is identification information for identifying a speaker. For example, the navigation device 10 may automatically generate the speaker ID. The speech feature vector is a feature vector that shows the characteristics of the speech emitted by that speaker. The speaker database may further store, associated with the speaker ID, at least one of the speaker's name and notification information destination information. Notification information destination information is information indicating the destination of the notification information. For example, the notification information destination information may be the caller's telephone number or email address.

[0032] The speech history database stores a history of speeches made by passengers of vehicle 30. The speech history database may store a history of only inappropriate words among those that have been spoken. For example, the speech history database may store, for each instance of speech, a non-command word and the date and time of use associated with the speaker ID. The speaker ID indicates the passenger who made the speech. The non-command word indicates a word spoken by the passenger that is not used to control the navigation device 10. The date and time of use indicates the date and time the word was spoken. The speech history database shown in Figure 3(b) shows that speaker A used "What's up?" twice and "You're in the way" once. On the other hand, speaker B is shown to have used "Aren't we there yet?" once.

[0033] The communication unit 13 controls communication with other devices via wireless communication. The communication unit 13 may include, for example, a network adapter and an antenna. For example, the communication unit 13 may be able to connect to a network such as the Internet via a base station (not shown). Also, for example, the communication unit 13 may be able to communicate with a portable terminal device.

[0034] The display 14 displays various information based on control by the control unit 11. The display 14 may be, for example, a liquid crystal display or an organic EL display. The speaker 15 outputs various sounds based on control by the control unit 11. The microphone 16 collects sounds such as the speech of passengers in the vehicle 30 and converts the collected sounds into audio signals. The microphone 16 outputs the audio signals to the control unit 11. The microphone 16 is an example of an audio input means. The input unit 17 is composed of, for example, a touch panel, buttons, switches, etc., and outputs operation signals indicating the operations performed by passengers in the vehicle 30 to the control unit 11.

[0035] The interface unit 18 is connected to the vehicle speed sensor 21, acceleration sensor 22, gyro sensor 23, GNSS sensor 24, and camera 25. The interface unit 18 performs interface processing between the navigation device 10 and the vehicle speed sensor 21 to camera 25. Each of the vehicle speed sensor 21 to camera 25 is mounted or installed on the vehicle 30.

[0036] The vehicle speed sensor 21 detects the vehicle speed 30 and outputs speed information indicating the detected speed to the interface unit 18. The acceleration sensor 22 detects the acceleration of the vehicle 30 and outputs acceleration information indicating the detected acceleration to the interface unit 18. The gyro sensor 23 detects the angular velocity of the vehicle 30 and outputs angular velocity information indicating the detected angular velocity to the interface unit 18. The GNSS sensor 24 uses GNSS (Global Navigation Satellite System) to receive signals transmitted from navigation satellites (not shown), calculates the position of the vehicle 30 based on these signals, and outputs position information (e.g., latitude and longitude) indicating the calculated position to the interface unit 18. The GNSS used may be, for example, GPS (Global Positioning System), or another system. The camera 25 is composed of, for example, a CCD (Charge Coupled Device) sensor, and captures images inside the vehicle 30 and outputs video data showing the captured images to the interface unit 18. This video data may be still images or video. The interior of vehicle 30 refers to the space within vehicle 30 that contains seats. For example, a camera 25 may be installed inside vehicle 30 so that it is possible to photograph the face of any person sitting in any of the seats.

[0037] [2. Functions of the Navigation System Control Unit] [2-1. Voice control] The control unit 11 controls the operation of the navigation device 10 according to the voice spoken by the passenger in the vehicle 30. The passenger inputs voice through the microphone 16. The control unit 11 generates a voice feature vector, for example, by performing a Fourier transform on the input voice signal. Based on the generated voice feature vector, the control unit 11 performs voice recognition to generate word information indicating the words spoken by the passenger. If the words spoken by the passenger match a command word stored in, for example, a voice command database, a genre database, or a location information database, or if it matches a combination of multiple command words stored in those databases, the control unit 11 may execute processing according to that command word or combination. For example, the passenger says the word "name" followed by the name of a facility "ABC". In response, the control unit 11 refers to the voice command database and recognizes that a search for a facility by name has been requested. Then, the control unit 11 searches the location information database for facilities with the name "ABC" among facilities located within a predetermined distance from the vehicle 30. The control unit 11 displays a list of the searched facilities on the display 14. When a passenger selects a facility from the list, the control unit 11 displays, for example, a map of the area around the selected facility on the display 14. If the passenger then says, "I'm going here," the control unit 11 recognizes that the selected facility is the destination by referring to the voice command database. The control unit 11 then searches for a route to the destination based on road data. For example, the control unit 11 may periodically calculate the current position of the vehicle 30 using the GNSS sensor 24, obtain location information indicating the current position from the GNSS sensor 24, and store it in the memory unit 12. The control unit 11 may then update or correct the location information stored in the memory unit 12 based on information output from the vehicle speed sensor 21, acceleration sensor 22, and gyro sensor 23. The control unit 11 may also use a route search algorithm such as Dijkstra's algorithm to search for a route from the current position to the destination. The control unit 11 then controls the display 14 and speaker 15 to provide route guidance to the destination.

[0038] The control unit 11 may interpret only the words uttered by the passenger following a specific operation or following a specific word as commands used by the passenger to control the navigation device 10. For example, the input unit 17 may include a speech button. Alternatively, the control unit 11 may display a speech button on the display 14. The control unit 11 may interpret only the words uttered by the passenger after pressing the speech button as commands. Alternatively, assume that the name of the navigation device 10 is "XYZ". The control unit 11 may interpret only the words uttered by the passenger after saying "XYZ" as commands.

[0039] [2-2. Speaker Registration] The control unit 11 registers passengers who have spoken in the vehicle 30, or passengers who may speak. This registration is performed by storing the information in the speaker database. For example, while the navigation device 10 is powered on, the microphone 16 may continuously collect sounds emitted in the vehicle 30. The control unit 11 generates a feature vector of the input sound using the microphone 16 and performs speaker recognition. For example, the control unit 11 compares the generated feature vector with the voice feature vector of a speaker already registered in the speaker database and calculates the similarity of the voice features. If the similarity is less than a predetermined value for any of the already registered speakers, the control unit 11 generates a new speaker ID. The control unit 11 may store the generated voice feature vector in the speaker database in association with the generated speaker ID.

[0040] The control unit 11 may register a passenger as a future speaker in response to an operation by the passenger. For example, the passenger selects to register as a speaker by operating the input unit 17 or by making a voice input to the microphone 16. In response, the control unit 11 may output a message via the speaker 15 prompting the passenger to utter a predetermined word. The passenger inputs voice into the microphone 16 according to the message. The control unit 11 generates a voice feature vector based on the input voice. The control unit 11 then stores the new speaker ID and voice feature vector in the speaker database. Furthermore, the passenger inputs a telephone number or email address as notification information recipient information for that passenger, for example by operating the input unit 17. The control unit 11 then stores the input notification information recipient information in the speaker database.

[0041] The control unit 11 may obtain notification information destination information from a predetermined system accessible via a network such as the Internet. The predetermined system may be, for example, a system that provides some kind of service to its members via the network. An example of a predetermined system is an SNS (Social Networking Service). For example, a passenger in vehicle 30 can access the system using a linked application by operating a terminal device carried by the passenger, and grant permission to the navigation device 10 to access information registered in the system. Members of such systems often register their email address or telephone number in the system.

[0042] [2-3. Reporting of inappropriate verbal habits] The control unit 11 acquires command words indicating words used to control a predetermined device and non-command words indicating words not used to control the predetermined device, based on the voice input using the microphone 16. The predetermined device may be, for example, a navigation device 10. For example, the control unit 11 may continuously collect voices emitted within the vehicle 30 using the microphone 16 while the power of the navigation device 10 is ON. The control unit 11 may then generate word information and determine whether the word information corresponds to a command word or a non-command word using the method described in Section 2-1.

[0043] The control unit 11 stores the acquired non-command words in the storage unit 12. For example, the control unit 11 may acquire the date and time when the voice was input from the microphone 16 as the usage date and time. The control unit 11 may then associate the non-command words with the usage date and time and store them in the speech history database. Here, the control unit 11 may perform speaker recognition based on the input voice. For example, the control unit 11 compares the voice feature vectors of each speaker stored in the speaker database with the feature vectors of the input voice and calculates the similarity of the voice features. The control unit 11 acquires the speaker ID of the speaker for whom a similarity of a predetermined value or higher has been calculated. The control unit 11 may then associate the acquired speaker ID with the non-command words and usage date and time and store them in the speech history database.

[0044] The control unit 11 identifies words that have been used more than a predetermined number of times, based on the non-command words stored in the memory unit 12. In other words, the control unit 11 identifies the passenger's catchphrases. For example, the control unit 11 may calculate the number of times a non-command word has been used by counting the number of usage dates and times associated with each non-command word or a specific non-command word stored in the speech history database. The predetermined number of times is two or more. For example, the predetermined number of times may be set in advance to a number of times that would determine the word to be a catchphrase. The control unit 11 may identify words that have been used more than a predetermined number of times for each speaker, who is distinguished based on the input voice. As mentioned above, non-command words and usage dates and times are stored in association with each speaker. Therefore, the control unit 11 can count the number of uses for each speaker according to this association. The control unit 11 may calculate the frequency of use of the non-command word as the number of uses. The control unit 11 may then identify words whose frequency of use exceeds a predetermined frequency as catchphrases. The frequency of use may be, for example, the number of times that non-control word is used per predetermined number of utterances. As mentioned above, the utterance history database stores the usage time for each non-command word used. Based on this usage time, the control unit 11 may calculate the usage frequency of each non-command word within the range of the most recent predetermined number of utterances. Alternatively, the frequency of use may be the number of times that non-control word is used per predetermined time. For example, the control unit 11 may calculate the usage frequency of each non-control word from before a predetermined time to the present time.

[0045] The control unit 11, if it determines that a word identified as being used more than a predetermined number of times corresponds to a word predetermined as an inappropriate word, will issue notification information indicating that word. In other words, the control unit 11 will notify the passenger of an inappropriate verbal tic. This notification will allow the speaker to become aware of their inappropriate verbal tic. It is also expected that this notification will serve as an opportunity to correct the verbal tic. The control unit 11 may determine whether the word identified as a verbal tic corresponds to an inappropriate word based on the inappropriate word database stored in the memory unit 12. The words predetermined as inappropriate words may be inappropriate words stored in the inappropriate word database. For an identified verbal tic to correspond to an inappropriate word may include the verbal tic and the inappropriate word being identical, or it may include the verbal tic and the inappropriate word being similar. In this case, similarity may include the verbal tic being a paraphrase of the inappropriate word to the extent that it does not change the nuance of the word. For example, the memory unit 12 may store a similar word database that associates words that are similar to each other.

[0046] The control unit 11 may, for example, compare each inappropriate word stored in the inappropriate word database with the word identified as a catchphrase. The control unit 11 may then determine that the catchphrase is an inappropriate word if it matches any of the inappropriate words. The control unit 11 may also obtain a similar word from the similar word database if it matches any of the words stored in the similar word database. The control unit 11 may then determine that the catchphrase is an inappropriate word if it finds that the retrieved similar word matches any of the inappropriate words stored in the inappropriate word database.

[0047] The control unit 11 may output the notification information as sound through the speaker 15, or display it on the display 14. Alternatively, the control unit 11 may transmit the notification information to a terminal device carried by the user, causing the terminal device to broadcast the notification information. An example of typical notification information is, "You should refrain from saying XXX." "XXX" is a specific inappropriate verbal tic. The control unit 11 may broadcast the notification information when an inappropriate verbal tic is identified, or when a passenger requests notification of an inappropriate verbal tic.

[0048] The control unit 11 may, depending on the situation in which the speaker is placed, perform either notification information via an output means connected to the navigation device 10, or transmit notification information to a terminal device different from the navigation device 10, which is used by the speaker who uttered the verbal tic, using an inappropriate word dictionary that indicates an inappropriate word corresponding to the verbal tic identified from among multiple inappropriate word dictionaries stored in the inappropriate word database. The control unit 11 may refer to a notification mode stored in the inappropriate word database in association with the inappropriate word dictionary that contains the inappropriate word corresponding to the identified verbal tic, and select either the output means connected to the navigation device 10 or the terminal device used by the speaker as the notification source. The output means connected to the navigation device 10 may or may not be provided by the navigation device 10 itself. This output means may include at least one of the display 14 and the speaker 15. The control unit 11 may transmit the notification information to the terminal device, for example, via email or short message. There are broadly two situations in which the speaker is placed. Specifically, the question is whether the person to whom the speaker's words are directed is in a location where they can perceive the notification information displayed by the navigation device 10 (display 14 or speaker 15). If the person to whom the speaker's words are directed cannot perceive the notification information, there is no problem with the navigation device 10 directly providing notification. In that case, the notification mode is set to "direct," and the control unit 11 may have the notification information displayed by the display 14 or speaker 15. On the other hand, if the person to whom the speaker's words are directed can perceive the notification information, and the navigation device 10 directly provides notification, the person to whom the words are directed will become aware of the speaker's inappropriate verbal tic. This could lead to awkwardness between the speaker and the person to whom the words are directed. However, if the information is sent to the speaker's terminal device, basically only the speaker can check the notification information. In that case, the notification mode is set to "speaker terminal," and the control unit 11 may send the notification information to the speaker's terminal device. By doing so, both the speaker and the listener can avoid having to experience such feelings.Specifically, the notification source may be controlled depending on whether or not the person to whom the inappropriate language is directed is assumed to be a passenger in vehicle 30. For example, the person to whom inappropriate language spoken by the driver is assumed to be outside vehicle 30. Therefore, the notification mode of the driver's inappropriate language dictionary may be set to "direct". The person to whom inappropriate language spoken by a passenger is assumed to be the driver of vehicle 30. Therefore, the notification mode of the passenger's inappropriate language dictionary may be set to "speaker terminal". The child to whom inappropriate language spoken by a parent is assumed to be a passenger in vehicle 30. Therefore, the notification mode of the parent's inappropriate language dictionary may be set to "speaker terminal".

[0049] Furthermore, the control unit 11 may also broadcast situation-appropriate notification information corresponding to an inappropriate word dictionary that indicates an inappropriate word corresponding to the word identified as a verbal tic, from among multiple inappropriate word dictionaries stored in the inappropriate word database. For example, the control unit 11 may change the message encouraging the speaker to refrain from using the inappropriate verbal tic for each inappropriate word dictionary (for example, "Please calm down a little," "That might be a bit harsh to a driver," "That's a bit overbearing to say to a child," etc.). By changing the notification information according to the situation in which the speaker is placed, it is expected that the speaker will be effectively made to stop using the inappropriate verbal tic. For example, the control unit 11 may broadcast a message as notification information according to a notification information template stored in the inappropriate word database, associated with an inappropriate word dictionary that contains the inappropriate word corresponding to the identified verbal tic.

[0050] Next, let's explain a specific example. As shown in Figure 4, person 40 is driving vehicle 30. At this time, due to poor traffic conditions, person 40 utters the words "You're in the way." 110 to another vehicle driving in front of vehicle 30. The frequency of person 40's use of word 110 has exceeded a predetermined frequency. Word 110 is stored in the driver's inappropriate word dictionary. The control unit 11 outputs notification information 120 from the speaker 15, according to the notification information template and notification mode associated with the driver's inappropriate word dictionary, saying, "Now, now, please calm down a little. Saying 'You're in the way.' is not good."

[0051] [3. Operation of the Navigation System] Next, an example of the operation of the navigation device 10 will be described. The control unit 11 executes the process shown in Figure 5 according to the various program codes contained in the program stored in the memory unit 12. The process shown in Figure 5 starts, for example, when the power of the navigation device 10 is turned ON.

[0052] First, the control unit 11 determines whether or not voice has been input via the microphone 16 (step S1). If voice has been input (step S1: YES), the control unit 11 generates a voice feature vector from the input voice signal and, based on the voice feature vector, generates word information indicating the spoken words through voice recognition (step S2). Next, the control unit 11 determines whether or not the words indicated by the word information correspond to a command word by referring to the voice command database, genre database, location information database, etc. (step S3). If the words indicated by the word information are a command word (step S3: YES), the control unit 11 executes processing according to that command word (step S4).

[0053] On the other hand, if the word indicated by the word information is not an imperative word (Step S3: NO), the control unit 11 compares the generated speech feature vector with the speech feature vector stored in the speaker database to obtain the speaker ID of the speaker who pronounced the word indicated by the word information from the speaker database (Step S5). Next, the control unit 11 determines, in association with the obtained speaker ID, whether the same word information as the generated word information is already stored in the speech history database (Step S6). If the same word information is not already stored in the speech history database (Step S6: NO), the control unit 11 stores the generated word information and the usage date and time indicating the current date and time in the speech history database in association with the speaker ID of the identified speaker (Step S7).

[0054] On the other hand, if the same word information is already stored in the speech history database (step S6: YES), the control unit 11 stores the usage date and time indicating the current date and time in the speech history database, associating the speaker ID of the identified speaker with the same word information as the generated word information (step S8). Next, the control unit 11 calculates the usage frequency of the word indicated by the generated word information based on the number of usage dates and times associated with the acquired speaker ID and the same word information as the generated word information (step S9). Next, the control unit 11 determines whether the calculated usage frequency is greater than a predetermined threshold (step S10). If the usage frequency is greater than the threshold (step S10: YES), the control unit 11 determines whether the word indicated by the generated word information is an inappropriate word by referring to the inappropriate word database (step S11).

[0055] If the word indicated by the word information is an inappropriate word (Step S11: YES), the control unit 11 identifies an inappropriate word dictionary stored in the inappropriate word database that indicates the word indicated by the generated word information as an inappropriate word. The control unit 11 obtains the notification information template and notification mode associated with the identified inappropriate word dictionary. Next, the control unit 11 determines whether the obtained notification mode is "direct" or not (Step S12). If the notification mode is "direct" (Step S12: YES), the control unit 11 embeds the generated word information into the obtained notification information template to generate notification information. Then, the control unit 11 converts the notification information into speech and outputs the notification information from the speaker 15 (Step S13). On the other hand, if the notification mode is "speaker terminal" (Step S12: NO), the control unit 11 embeds the generated word information into the notification information template to generate notification information. Then, the control unit 11 uses the notification information destination information stored in the speaker database in association with the acquired speaker ID to send the notification information to the speaker's terminal device (step S14).

[0056] If the frequency of use is below the threshold (step S10: NO), or if the word indicated by the word information is not an inappropriate word (step S11: NO), or after steps S4, S7, S13, or S14, the control unit 11 determines whether the power to the navigation device 10 has been turned OFF (step S15). If the power has not been turned OFF (step S15: NO), the process proceeds to step S1. If the power has been turned OFF (step S15: YES), the process shown in Figure 5 ends.

[0057] As described above, the navigation device 10 according to the embodiment may acquire, based on the voice input using the microphone 16, command words indicating words used to control the navigation device 10 and non-command words indicating words not used to control the navigation device 10. The navigation device 10 may also store the acquired non-command words in the storage unit 12. The navigation device 10 may also identify words that have been used more than a predetermined number of times based on the non-command words stored in the storage unit 12. Furthermore, if the identified word corresponds to a word predetermined as an inappropriate word, the navigation device 10 may announce notification information indicating that identified word. In this case, excluding words used to control the navigation device 10, it becomes possible to make the speaker aware of any inappropriate verbal habits.

[0058] Furthermore, the navigation device 10 may identify words that are used more than a predetermined number of times for each speaker, based on the voice input using the microphone 16. In this case, inappropriate verbal habits can be reported for each speaker.

[0059] Furthermore, the navigation device 10 may determine whether the identified word corresponds to an inappropriate word based on the inappropriate word dictionary stored in the memory unit 12. Here, the memory unit 12 may store multiple inappropriate word dictionaries, each of which indicates an inappropriate word according to a set of different situations in which the speaker finds themselves. The navigation device 10 may then perform either notification information via an output means connected to the navigation device 10, or transmit notification information to a terminal device different from the navigation device 10, used by the speaker who uttered the verbal tic, depending on the situation corresponding to the inappropriate word dictionary that indicates an inappropriate word corresponding to the identified word. In this case, depending on the situation in which the speaker finds themselves, it is possible to control whether or not it is acceptable for others to know that the speaker has an inappropriate verbal tic.

[0060] Furthermore, the navigation device 10 may also provide situation-appropriate notification information corresponding to an inappropriate word dictionary that indicates an inappropriate word equivalent to the identified word, selected from among multiple inappropriate word dictionaries stored in the memory unit 12. In this case, the notification information can be tailored to the speaker's situation, prompting the speaker to refrain from using their verbal tic.

[0061] [4. Variant] The control unit 11 may identify the situation in which the speaking passenger is located. The control unit 11 may then calculate the number of times each spoken word is used for each situation in which the speaker is located, and determine whether or not that word is a catchphrase. The control unit 11 may then use one of several inappropriate word dictionaries, specifically the one corresponding to that situation, to determine whether or not the catchphrase is inappropriate for each identified situation. For example, the phrase "I thought there would be a traffic jam" is inappropriate when used by a passenger towards the driver, but it is not necessarily inappropriate when used by the driver themselves. The control unit 11 may, for example, have the camera 25 take pictures inside the vehicle 30 while the navigation device 10 is powered on. The control unit 11 may then identify the situation in which the speaker is located based on the images taken by the camera 25. For example, the control unit 11 may identify the passengers in the vehicle 30 using algorithms for general object detection and human posture estimation. The control unit 11 may identify a passenger who is moving their mouth at the same time as the voice input as the speaker. The control unit 11 may determine whether the speaker is the driver or a passenger based on the speaker's position in the captured video. Specifically, the control unit 11 may determine that the speaker is the driver if the speaker is sitting in the driver's seat. On the other hand, the control unit 11 may determine that the speaker is a passenger if the speaker is sitting in a seat other than the driver's seat. The control unit 11 may also determine whether both adults and children are in the vehicle 30 based on the captured video. For example, the control unit 11 may determine whether a person is an adult or a child based on the person's build or face size in the video, or the position of their eyes, nose, and mouth relative to their face. Alternatively, the control unit 11 may determine whether the speaker is an adult or a child based on the feature vector of the input voice. Furthermore, in a situation where both an adult and a child are in the vehicle 30, the control unit 11 may determine that the parent (or simply the adult) spoke to the child if the adult speaks.

[0062] The control unit 11 may determine the legitimacy of the driver's driving or judgment if the words spoken by a passenger in the vehicle 30 are inappropriate. If the driving or judgment is deemed legitimacy, the control unit 11 may also include information indicating this legitimacy in the notification information. For example, the control unit 11 may search for one or more appropriate routes from the departure point to the destination or current location if the vehicle 30 were to start moving at its actual departure time. An appropriate route is one that takes relatively little time to travel. In this case, the control unit 11 may search for routes considering traffic congestion information, road closure information, etc. If the current route is included in the appropriate routes, the control unit 11 may determine that the driver's route selection was legitimacy. An example of notification information in this case would be, "XXX might be a bit harsh on the driver. If you left at that time, that route was the best option." This is expected to discourage passengers from complaining or expressing dissatisfaction.

[0063] When the control unit 11 broadcasts notification information, it may identify the speaker's tone of voice. For example, the control unit 11 may identify the tone of voice based on the speaker's volume, tone, tempo, speed, intonation, etc. If the identified tone of voice is inappropriate, the control unit 11 may also include information about that tone of voice in the notification information. An inappropriate tone of voice may include, for example, a tone of voice that would offend the listener. An example of an inappropriate tone of voice is a strong tone of voice. A strong tone of voice may be, for example, a voice volume greater than a predetermined value. For example, if a parent uses an inappropriate tone of voice towards a child, an example of notification information may be, "XXX's way of speaking to the child was a little overbearing. Their tone of voice was also a little strong." This is expected to soften the speaker's tone of voice.

[0064] In the embodiment described above, the control unit 11 stored information for all non-command words in the speech history database and determined whether a non-command word whose usage count exceeded a predetermined number of times was an inappropriate word. However, the control unit 11 may store information for only inappropriate non-command words in the speech history database. Furthermore, the control unit 11 may issue notification information if the usage count of an inappropriate word exceeds a predetermined number of times.

[0065] The notification information may include both the identified inappropriate word and an appropriate word with the same meaning. This allows the speaker to learn the correct way to say it, and is therefore expected to help the speaker correct their inappropriate speech patterns.

[0066] In the embodiment described above, the notification device D according to the embodiment and the voice-controlled device were the same. However, the notification device D and the voice-controlled device may be different. For example, the notification device D may be applied to a server device that can communicate with the navigation device 10 via a network. In this case, the navigation device 10 may be a voice-controlled device. At least one of the inappropriate word database, the speaker database, and the speech history database may be stored in the server device. The server device may, for example, acquire voice data input via the microphone 16 from the navigation device 10 and execute the processing described in the embodiment. The server device may also transmit a command to the navigation device 10, causing the navigation device 10 to broadcast notification information.

[0067] Furthermore, notification device D may be applied to other devices. For example, notification device D may be applied to a smart speaker. Alternatively, notification device D may be applied to terminal devices such as personal computers. For example, notification device D may be applied to terminal devices used by call center operators. In this case, inappropriate language may include, for example, rude language directed at the customer on the phone. Examples of such language include "Hello," "Just in case," "I can't," "Absolutely not," "Thank you for your help," and "Your voice is too quiet, I can't hear you." An example of notification information is, "'Your voice is too quiet, I can't hear you' is a rude way of putting the blame on the customer. Try to use phrases that confirm whether you can hear each other, such as, 'The phone connection seems a little poor, can you hear me?'" Notification information may be given after the phone call has ended. [Explanation of symbols]

[0068] 1 Acquisition method 2. Memory control means 3 Specific means 4. Notification control means 5 Microphone 6 Memory means D. Notification device 10 Navigation System 11 Control Unit 12 Storage section 13 Communications Department 14 displays 15 speakers 16 Microphones 17 Input section 18 Interface section 21. Vehicle speed sensor 22 Accelerometer 23 Gyroscope Sensor 24 GNSS sensors 25 Cameras 30 vehicles

Claims

1. An acquisition means that acquires non-control word information indicating words not used to control a predetermined device, based on audio input using an audio input means, A memory control means for storing the acquired non-control word information in a memory means, Based on the stored non-control word information, a means for identifying words that are used more than a predetermined number of times, If the identified word corresponds to a word predetermined as an inappropriate word, a notification control means causes notification information indicating the identified word to be broadcast. Equipped with, The aforementioned inappropriate words include at least one of the following: rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms. The aforementioned storage means stores inappropriate word information that indicates words that are inappropriate depending on the situation in which the speaker is placed. The notification control means, based on inappropriate word information indicating an inappropriate word corresponding to the identified word, (a) The notification information is broadcast by an output means connected to the predetermined device, or (b) A notification device characterized by transmitting the notification information to a terminal device different from the predetermined device, which is used by the speaker.

2. An acquisition means that acquires non-control word information indicating words not used to control a predetermined device, based on audio input using an audio input means, A memory control means for storing the acquired non-control word information in a memory means, Based on the stored non-control word information, a means for identifying words that are used more than a predetermined number of times, If the identified word corresponds to a word predetermined as an inappropriate word, a notification control means causes notification information indicating the identified word to be broadcast. Equipped with, The aforementioned inappropriate words include at least one of the following: rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms. The aforementioned storage means stores inappropriate word information that indicates words that are inappropriate depending on the situation in which the speaker is placed. The notification control means is characterized by causing the notification device to notify the notification information according to the situation based on inappropriate word information indicating an inappropriate word corresponding to the identified word.

3. The notification device according to claim 1 or 2, characterized in that the identifying means identifies a word whose usage count is greater than a predetermined number of times for each speaker distinguished based on the input voice.

4. In a notification method performed by a computer, An acquisition step of acquiring non-control word information that indicates words not used to control a predetermined device, based on the sound input using an audio input means, A memory control step involves storing the acquired non-control word information in a memory means, Based on the stored non-control word information, the identification step involves identifying words that are used more than a predetermined number of times. If the identified word corresponds to a word predetermined as an inappropriate word, a notification control step is performed to cause notification information indicating the identified word to be broadcast. Includes, The aforementioned inappropriate words include at least one of the following: rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms. The aforementioned storage means stores inappropriate word information that indicates words that are inappropriate depending on the situation in which the speaker is placed. The notification control step is performed based on inappropriate word information indicating an inappropriate word corresponding to the identified word, (a) The notification information is broadcast by an output means connected to the predetermined device, or (b) A notification method characterized by transmitting the notification information to a terminal device different from the predetermined device, which is used by the speaker.

5. In a notification method performed by a computer, An acquisition step of acquiring non-control word information that indicates words not used to control a predetermined device, based on the sound input using an audio input means, A memory control step involves storing the acquired non-control word information in a memory means, Based on the stored non-control word information, the identification step involves identifying words that are used more than a predetermined number of times. If the identified word corresponds to a word predetermined as an inappropriate word, a notification control step is performed to cause notification information indicating the identified word to be broadcast. Includes, The aforementioned inappropriate words include at least one of the following: rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms. The aforementioned storage means stores inappropriate word information that indicates words that are inappropriate depending on the situation in which the speaker is placed. The notification control step is characterized by causing the notification information corresponding to the situation to be notified based on inappropriate word information indicating an inappropriate word corresponding to the identified word.

6. Computers, An acquisition means that acquires non-control word information indicating words not used to control a predetermined device, based on audio input using an audio input means, A memory control means for storing the acquired non-control word information in a memory means, Based on the stored non-control word information, a means for identifying words that are used more than a predetermined number of times, If the identified word corresponds to a word predetermined as an inappropriate word, a notification control means causes notification information indicating the identified word to be broadcast. To make it function as, The aforementioned inappropriate words include at least one of the following: rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms. The aforementioned storage means stores inappropriate word information that indicates words that are inappropriate depending on the situation in which the speaker is placed. The notification control means, based on inappropriate word information indicating an inappropriate word corresponding to the identified word, (a) The notification information is broadcast by an output means connected to the predetermined device, or (b) A notification program characterized by causing the notification information to be transmitted to a terminal device different from the predetermined device, which is used by the speaker.

7. Computers, An acquisition means that acquires non-control word information indicating words not used to control a predetermined device, based on audio input using an audio input means, A memory control means for storing the acquired non-control word information in a memory means, Based on the stored non-control word information, a means for identifying words that are used more than a predetermined number of times, If the identified word corresponds to a word predetermined as an inappropriate word, a notification control means causes notification information indicating the identified word to be broadcast. To make it function as, The aforementioned inappropriate words include at least one of the following: rude words, offensive words, dirty words, words that give a negative impression, slang, and discriminatory terms. The aforementioned storage means stores inappropriate word information that indicates words that are inappropriate depending on the situation in which the speaker is placed. The notification control means is characterized by causing the notification program to notify the notification information according to the situation based on inappropriate word information indicating an inappropriate word corresponding to the identified word.

8. A recording medium characterized by having the notification program described in claim 6 or 7 recorded in a computer-readable format.

Images