Control unit by means of voice

Abstract

To solve such problems that, from the viewpoint of preventing frauds, some conventional phones display a caller number on a display section, notify, to a caller, that phone conversation is being recorded, or notify the danger of a communication partner by converting contents talked by the partner into a text in real time to collate the text with words that has a high possibility of a fraud, but, even if information on a danger is provided by a method mentioned above, a determination for whether or not subsequent conversation is to be continued is entrusted to a receiver, so that, if explanation of the caller is clever or a partner who is talking claims to be a relative or friend, a risk of an occurrence of an error in determination for whether or not the conversation is to be continued becomes high.SOLUTION: There is provided a control unit in which, a determination for whether or not conversation is to be continued is entrusted to the control unit while a countermeasure for a case where errors in determination to forcibly cut off a call occur is taken, so that, regardless of the kind of move made by a caller, a communication can be mechanically terminated and a fraud damage can be prevented.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a voice-activated control device that distinguishes the contents of a telephone conversation and performs an action according to the contents. [Background technology]

[0002] In recent years, the number of victims of fraudulent bank transfer scams, which aim to extract money from people, has been increasing. As a countermeasure, there are methods such as displaying the caller's number on the display and notifying the other party that a telephone conversation is being recorded.

[0003] There are also devices, such as those described in Patent Document 1, that convert what the other person is saying into text in real time and compare it with a list of dangerous words often used in scams to notify the recipient of the danger of the other party, and devices, such as those described in Patent Document 3, that simultaneously compare the converted text with the other party's voice. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2007-228384 A [Patent Document 2] JP 2011-151451 A [Patent Document 3] JP 2008-210085 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, with the above method, only information about the dangers is provided, and the decision of whether to continue the conversation is left to the recipient. Therefore, there is a problem in that if the caller is skillful in his / her explanation or the person he / she is speaking to identifies themselves as a relative or friend, there is a high possibility that the recipient will make a mistake in deciding whether to continue the conversation.

[0006] In recent years, there have also been cases of people using voice synthesis to imitate the voices of relatives and friends, making it more difficult to distinguish between them.

[0007] SUMMARY OF THE PRESENT DISCLOSURE ... In order to solve the above problems, an object of the present invention is to provide a voice-based control device that mechanically ends communication by determining whether to continue the conversation regardless of the caller's response. [Means for solving the problem]

[0008] In order to solve the above problems, the voice-based control device according to the present invention comprises communication means for communicating with a communication line, received voice control means for receiving received voice from a telephone and controlling a signal, transmitted voice control means for controlling transmitted voice sent to the telephone and controlling the hook of the telephone, voice generation means for converting received voice signals into voice audible to humans, voice-to-text means for recognizing received voice and converting it into text, discrimination means for making a judgment using information obtained from the voice-to-text means, pre-registered text information, and new text information from the outside obtained using the communication means, or for transmitting the information obtained during discrimination to an external server using the communication means when the text information is encountered for the first time and is considered useful for discrimination, message transmission means for transmitting a specific message at the instruction of the discrimination means, voice input means for converting voice uttered by a person into an electrical signal, recording means for recording information referred to by the judgment means during judgment, information obtained by the communication means, and a telephone number to be notified to the caller, and input / output means for inputting set values ​​and outputting information to the outside. Effect of the Invention

[0009] The above voice-controlled device has a countermeasure in case of an error in the decision to forcibly disconnect the call, and leaves the decision of whether to continue the conversation to the discrimination means, so that the communication can be mechanically terminated regardless of the caller's response.

[0010] Conversely, it also has the effect of allowing you to confirm whether someone is trustworthy by using a password word list.

[0011] Furthermore, in case the decision to forcibly cut off the call turns out to be a mistake, the system has the ability to notify the caller of the contact information of a person who can make the appropriate decision when the communication is ended, and to maintain a password word list that has been pre-registered for trusted callers, and to unconditionally allow further conversation when one of these words is uttered.

[0012] Furthermore, since the discrimination means provides and receives information between users of this device via the communication means, the users always share the latest dangerous word list, thereby reducing damage caused by delays in information. [Brief description of the drawings]

[0013] [Figure 1] FIG. 1 is a diagram showing a configuration of an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram showing a configuration of an embodiment of the present invention, and is an image diagram of a dangerous word list registered in a storage means 9 in a memory. [Diagram 3] FIG. 2 is a diagram showing a configuration of an embodiment of the present invention, and is an image diagram of a password word list registered in a storage means 9 in a memory. [Figure 4] FIG. 1 is a diagram showing a configuration of an embodiment of the present invention, in which the determination means is provided with a function of analyzing a received voice and determining the person who uttered the voice. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail with reference to the drawings. EXAMPLES

[0015] Fig. 1 is a diagram for explaining the first embodiment. As shown in Fig. 1, this voice-based control device includes a communication means 1, a received voice control means 2, a voice generation means 3, a voice-to-text means 4, a discrimination means 5, a transmitted voice control means 6, a message transmission means 7, a voice input means 8, a recording means 9, and an input / output means 10.

[0016] The communication lines described here are assumed to be telephone switching networks or the Internet, and when an Internet environment is not available, communication is carried out using telephone lines.

[0017] Also, while FIG. 1 shows a method of converting speech into text using the communication means 1, this speech-to-text conversion means 4 may be completed within this device without using the communication means 1.

[0018] First, when the present device is installed, the determination means 5 uses the communication means 1 to obtain a dangerous word list from the data server 11 , and the list is recorded in the storage means 9 .

[0019] The image of the dangerous word list in memory is shown in Figure 2, and words are registered in sequence. The score shown here indicates the level of danger, and the more frequently a word is used when a fraud occurs, the higher the number will be.

[0020] The password word list is either manually input directly into the device using the input / output means 10 or input by a relative or the like using the communication means 1 and then recorded in the storage means 9.

[0021] The image of the password word list in memory is shown in Figure 3, and words are registered in sequence. Note that there does not necessarily have to be multiple password words.

[0022] Furthermore, the telephone number to be notified to the caller when the call is forcibly disconnected is also manually entered directly into the device using the input / output means 10, or entered by a relative or the like using the communication means 1, and then recorded in the memory means 9.

[0023] The upper limit value of the score memory which determines the degree of danger of dangerous words is also similarly entered manually into the device using the input / output means 10, or entered by a relative or the like using the communication means 1, and recorded in the memory means 9.

[0024] For the sake of explanation, we will assume that the dangerous word list is as shown in Figure 2, the password word list is as shown in Figure 3, the phone number notified to the caller when a call is forcibly disconnected is the phone number of a relative, and the upper limit of the score memory for judging the degree of danger of dangerous words is set to 20.

[0025] The device then goes into standby mode until a call is received.

[0026] When a call is received, the start and end of the ringing sound or the voice signal of the caller or the call recipient is detected, and the discrimination means 5 recognizes that a conversation has started and resets the contents of a score memory for judging the degree of danger of the conversation to 0. As for the voice, the sender's voice is transmitted to the receiver's ears via the voice generating means 3, and the receiver's voice is transmitted to the sender via the voice input means 8.

[0027] When monitoring by this device begins, the caller's voice is converted into text by the voice-to-text conversion means 4 and sent to the discrimination means 5 in sequence.

[0028] The discrimination means 5 monitors the text and checks each word against the dangerous word list and the password word list.

[0029] If the result of the comparison is that the word belongs to the dangerous word list, the score described together with the dangerous word is obtained and added to a score memory that determines the degree of danger of the conversation, and a check is made to see if the score has reached an upper limit.

[0030] Similarly, if the matched result is a word that belongs to the password word list, the system immediately stops converting the caller's voice into text and goes into standby mode.

[0031] When a score memory, which is used to determine the danger level of a conversation by addition, reaches an upper limit value, the determining means 5 instructs the message transmitting means 7 to transmit the next message.

[0032] "I'm sorry. I'm having trouble understanding what you're saying so I'll hang up the phone for now. If it's urgent or you absolutely need to get in touch, please call XXX again." In addition, the telephone number recorded in the storage means 9 is set as voice in the portion of XXX. Furthermore, the content of the conversation is free and does not necessarily have to match the above content.

[0033] Thereafter, the call is forcibly disconnected by the telephone hook drive of the transmission voice control means 6, and the recipient is informed by the message transmission means 7 that "The call was disconnected because the content of the call was suspected to be fraud." The content of the conversation is free and does not necessarily have to match the above content.

[0034] Incidentally, the transmission voice control means 6 does not need to have a part for driving the telephone hook if the telephone call can be disconnected by another method.

[0035] Furthermore, the discrimination means 5 extracts words that are not on the dangerous word list but appear frequently, and registers the information in the data server 11 using the communication means 1.

[0036] Finally, the telephone hook drive of the transmission voice control means 6 is released, and the state returns to the standby state.

[0037] By the way, there is a concern that the caller may talk to the recipient to find out the password for this device, so as shown in Figure 2, words that are thought to be used to find out the password are given a high score, and if they are used, the call is immediately cut off. EXAMPLES

[0038] Fig. 4 is a diagram for explaining the second embodiment. As shown in Fig. 4, this voice-based control device is obtained by adding a voice analysis means 13 to the first embodiment.

[0039] It should be noted that the voice analysis means 13 may not use the communication means 1, similar to the voice-to-text conversion means 4.

[0040] The difference between the second embodiment and the first embodiment is in the password matching, which reduces the possibility of the concern shown at the end of the first embodiment occurring.

[0041] In the first embodiment, a password word is collated to identify trustworthy senders, but as mentioned above, there is a possibility that the word itself may be transmitted to someone committing fraud, so this method is insufficient as a means of identification.

[0042] The second embodiment performs password matching in the same manner as the first embodiment, but also analyzes the spoken voice to verify whether the characteristics of the sound itself match those of a trusted caller that has been registered in advance.

[0043] In this case, matching techniques such as voiceprints are used for matching. EXAMPLES

[0044] The third embodiment is not intended to prevent fraud itself, but is used as a tool to confirm that the sender is a trustworthy sender.

[0045] As explained in the second embodiment, password matching and voice analysis matching are useful tools for caller verification.

[0046] In the second embodiment, when the caller is not a trustworthy caller, an action is taken to block the call, whereas in this embodiment, when the caller is a trustworthy caller, an output to that effect is sent to the input / output means 10 to notify the caller of caller verification.

[0047] Generally, when communicating over the phone, the recipient identifies the other person by their voice and speaking style, but it is already possible to synthesize voice using generative AI and other technologies, which increases the possibility of phone spoofing.

[0048] The third embodiment can easily identify the other party through a dual configuration of password matching and voice analysis matching. EXAMPLES

[0049] The fourth embodiment is a case where a service is provided regarding a telephone number that is notified to a caller when a call is forcibly disconnected.

[0050] This device recommends calling another phone number as a countermeasure in case there is an error in the decision to forcibly disconnect the call. While it is unlikely that calls will be made if the call itself is a scam, they may be made in other cases and may result in costs for recipients of the targeted phone number.

[0051] In this embodiment, the telephone receiving service is operated as a business, and the recipient of the telephone receiving service talks with the caller and responds appropriately. [Explanation of symbols]

[0052] 1. Means of communication 2. Receiving audio control means 3. Sound Generation Means 4. Speech-to-text conversion methods 5 Discrimination means 6. Transmitting voice control means 7. Means of sending messages 8. Voice input means 9. Recording means 10 Input / output means 11 Data Server 12 Telephone 13 Voice Analysis Methods

Claims

1. a voice control device comprising: a communication means for communicating with a communication line; an incoming voice control means for receiving an incoming voice from a telephone and controlling a signal; an outgoing voice control means for controlling the outgoing voice sent to the telephone and controlling the telephone hook; a voice generating means for converting an incoming voice signal into a voice audible to humans; a voice text means for recognizing and converting the incoming voice into text; a discrimination means for making a judgment using information obtained from the voice text means, pre-registered text information, and new text information from the outside obtained using the communication means, or for sending the text information obtained during discrimination to an external server using the communication means when the text information is encountered for the first time and is considered to be useful for discrimination; a message transmitting means for transmitting a specific message at the instruction of the discrimination means; a voice input means for converting a voice spoken by a human into an electrical signal; a recording means for recording information referred to by the judgment means during judgment, information obtained by the communication means, and a telephone number to be notified to the caller; and an input / output means for inputting setting values ​​and outputting information to the outside.

2. 2. A voice control device according to claim 1, wherein said discrimination means is provided with a path for inputting information from voice analysis means for analyzing a received voice and discriminating the person who has uttered the voice.

Images