TWI835304B - Ai voice control banking transaction system - Google Patents

Abstract

An artificial intelligence voice control banking transaction system includes an automatic teller machine, a computing device, a memory and a voice receiving device. The computing device is used for authentication operation and command confirmation operation, the memory is used for storing two voice messages, as well as the voice recognition model, registered identity data and command database, and the voice receiving device is used for receiving and transmitting the two voice messages to the memory. The authentication operation includes retrieving the pending identity data in the first voice message by using the voice recognition model, and comparing and confirming that the pending identity data is consistent with the registered identity data, and the command confirmation operation includes retrieving a second voice message through the voice recognition model pending command data in the command database, and an execution command data corresponding to the plurality of execution command data in the command database by comparing and judging the pending command data

Description

Artificial intelligence voice-controlled banking transaction system

The invention relates to a bank transaction system, and in particular to an artificial intelligence voice-activated bank transaction system.

Current bank transaction systems, such as automated teller machines (ATMs), mostly have transaction interfaces in the form of buttons or panels, which causes certain inconveniences in certain situations or for specific groups (such as those with visual impairment and mobility issues). sex.

In some systems, voice functions are often assistive and need to be paired with screen display and touch braille. Therefore, for certain users, using assistive ATMs is often not as convenient as using online banking through mobile phones.

In view of the above, the present invention provides an artificial intelligence voice-activated bank transaction system.

An artificial intelligence voice-activated banking transaction system according to an embodiment of the present invention includes an automatic teller machine, a computing device connected to the automatic teller machine, a memory connected to the computing device, and a voice receiving device connected to the memory. The automatic teller machine is used to allow users to handle bank counter services by themselves. The computing device is used to perform an identity verification operation and an instruction confirmation operation on the user. The memory is used to store a first and second voice message and also stores voice recognition. model, a registered identity information of the user and a command database, and the voice receiving device is used to receive The user's first and second voice messages are transmitted to the memory. The identity verification operation performed by the computing device includes: retrieving a pending identity data contained in the first voice message through a speech recognition model, and comparing the pending identity data with the registered identity data and confirming that the pending identity data matches the registered identity data. , and the instruction confirmation operation executed by the computing device includes: retrieving a pending instruction data contained in a second voice message through a speech recognition model, and comparing the pending instruction data with multiple execution instruction data in the instruction database, To determine an execution instruction data corresponding to the pending instruction data.

With the above structure, the artificial intelligence voice-activated banking transaction system disclosed in this case can judge the content of the received voice messages through a memory and computing device that stores an artificial intelligence speech recognition model, so as to safely and effectively execute identity verification and instructions. Confirm function. Users only need to state their identity or account information in front of the ATM to complete the verification, and then verbally express the banking services they want to perform. In this way, general banking services can be completed completely through voice control, and various It is a convenient measure for all ethnic groups.

The above description of the present disclosure and the following description of the embodiments are used to demonstrate and explain the spirit and principles of the present invention, and to provide further explanation of the patent application scope of the present invention.

1: Artificial intelligence voice-controlled banking transaction system

10:ATM

20:Computing device

30,30’,30”,30''’: memory

31: Speech recognition model

32:Register identity information

33: Command database

34: Facial feature detection model

35: Log in facial features

36: Lip shape recognition model

37: First trigger message

38: Second trigger message

40: Voice receiving device

50:Image capture device

C:user

S40: Identity verification operation

S400~S402,S500~S503,S600~S604,S700~S703,

Figure 1a is a block diagram of an artificial intelligence voice-activated banking transaction system according to an embodiment of the present invention.

Figure 1b is a block diagram illustrating a usage scenario of the artificial intelligence voice-activated banking transaction system according to an embodiment of the present invention.

FIG. 1c is a block diagram illustrating the memory of the artificial intelligence voice-activated banking transaction system receiving voice messages according to an embodiment of the present invention.

FIG. 2a is an operation flow chart of an artificial intelligence voice-activated banking transaction system performing identity verification according to an embodiment of the present invention.

Figure 2b is an operation flow chart of an artificial intelligence voice-activated banking transaction system executing instruction confirmation according to an embodiment of the present invention.

FIG. 3 is another operational flow chart of the artificial intelligence voice-activated banking transaction system performing identity verification according to an embodiment of the present invention.

FIG. 4 is another operation flow chart of the artificial intelligence voice-activated banking transaction system performing operation confirmation according to an embodiment of the present invention.

Figure 5a is a block diagram of an artificial intelligence voice-activated banking transaction system according to another embodiment of the present invention.

Figure 5b is a block diagram illustrating a usage scenario of the artificial intelligence voice-activated banking transaction system according to another embodiment of the present invention.

FIG. 5c is a block diagram illustrating the memory of the artificial intelligence voice-activated banking transaction system receiving voice messages and facial image data according to another embodiment of the present invention.

FIG. 6 is an operation flow chart of an artificial intelligence voice-activated banking transaction system performing facial recognition for identity verification according to another embodiment of the present invention.

FIG. 7 is a block diagram illustrating another memory of the artificial intelligence voice-activated banking transaction system receiving messages and data according to another embodiment of the present invention.

FIG. 8 is an operation flow chart of an artificial intelligence voice-activated banking transaction system executing lip recognition for instruction confirmation according to yet another embodiment of the present invention.

FIG. 9 is a block diagram illustrating the usage scenario of the artificial intelligence voice-activated banking transaction system according to other embodiments of the present invention.

FIG. 10 is a block diagram illustrating other memories of the artificial intelligence voice-activated banking transaction system receiving messages and data according to other embodiments of the present invention.

FIG. 11 is a block flow chart of the artificial intelligence voice-activated banking transaction system when triggered according to other embodiments of the present invention.

The detailed features and advantages of the present invention are described in detail below in the implementation mode. The content is sufficient to enable anyone skilled in the relevant art to understand the technical content of the present invention and implement it according to the content disclosed in this specification, the patent scope and the drawings. , anyone familiar with the relevant art can easily understand the relevant objectives and advantages of the present invention. The following examples further illustrate the aspects of the present invention in detail, but do not limit the scope of the present invention in any way.

Please refer to Figures 1a to 1c. Figure 1a is a block diagram of an artificial intelligence voice-activated banking transaction system according to an embodiment of the present invention. Figure 1b is a block diagram of an artificial intelligence voice-activated banking transaction system according to an embodiment of the present invention. A block diagram of a usage scenario, FIG. 1c is a block diagram of the memory of the artificial intelligence voice-activated banking transaction system according to an embodiment of the present invention receiving voice messages.

As shown in Figures 1a and 1b, the artificial intelligence voice-activated banking transaction system 1 includes an automatic teller machine 10, a computing device 20 connected with signals to the ATM 10, a memory 30 connected with signals to the computing device 20, and a memory 30 connected with signals Voice receiving device 40. Automatic cabinet The employee machine 10 is used to allow the user C to handle bank counter services by himself, such as, but not limited to, an automated teller machine (ATM) used by the general public. The computing device 20 is used to perform an identity verification operation and an instruction confirmation operation on the user C, that is, to verify the identity of the user C and determine the instructions that the user wants to issue. The computing device 20 can be a variety of processes with computing capabilities. device. The memory 30 is used to store a first and a second voice message, and stores a voice recognition model 31, a registered identity information 32 of the user C, and an instruction database 33. The memory 30 can be a variety of devices with the ability to store data. Capacity storage device. The voice receiving device 40 is used to receive the first and second voice messages of the user C and transmit them to the memory 30. The voice receiving device 40 can be a microphone and can be disposed adjacent to the automatic teller machine 10 to receive the voice messages from the user C. Voice.

Please refer to FIG. 1c together. The voice receiving device 40 transmits the first voice message and the second voice message to the memory 30 in the form of signal transmission. The above two voice messages will be analyzed and judged by the computing device 20 through the speech recognition model 31, and the first voice message and the second voice message respectively correspond to the registered identity data 32 and the command database 33, which will be described later. It should be noted that the connections between different blocks in Figures 1a to 1c can be solid lines or dotted lines. Specifically, the signal connections between different blocks can be connected through physical signal lines or through wireless communication. There is no restriction in this case. On the other hand, as can be understood by those with ordinary knowledge of the case, except for the voice receiving device 40 in FIG. 1 b which needs to be disposed adjacent to the ATM 10 to meet the favorable conditions for physical sound wave transmission, the other blocks can be disposed remotely in other locations. , for example, the computing device 20 can be a central processing unit, and the memory 30 can be a cloud database, etc. Of course, in some embodiments, the computing device 20 and the memory 30 may be disposed adjacent to the ATM 10 Recently, in order to achieve efficient edge computing and prevent data interception, this part will be described later and should not be a restriction in this case.

The speech recognition model 31 in this example can be a neural network pre-trained by a deep learning method. Specifically, the speech recognition model 31 can include automatic speech recognition (Automatic Speech Recognition, ASR) software, so that in theory, the calculation The device 20 can perform the following steps through automatic speech recognition software to complete the speech recognition function of this case. 1. The user sends a voice message to the voice receiving device. 2. Convert the sound message into a sound wave signal. 3. The filter in the voice receiving device filters out the noise in the acoustic signal. 4. Decompose the filtered sound wave signal into multiple groups of phonemes (phonemes refer to the basic sound blocks that make up the sounds of language. In English, there are 44 phonemes such as "wh", "th" and "t". ", etc., while the Chinese system lacks unified standards and has no fixed number). 5. Each phoneme can form a list and can be analyzed statistically in sequence. 6. The semantic meaning of a passage can be understood through automatic speech recognition software.

The speech recognition model 31 based on the above-mentioned automatic speech recognition can further have two main variations, namely a direct dialogue (Directed Dialogue) model and a natural language processing (NLP) model. In this article, the direct dialogue model will be mainly used as an example, but in other embodiments, a natural language processing model can also be used, and this case is not limited thereto. The so-called direct dialogue model refers to a specific vocabulary selection that can directly provide the user with a certain range, making it easier for the machine to determine semantic meaning than using a natural language processing model.

Based on the above-mentioned artificial intelligence voice-activated banking transaction system 1 shown in Figures 1a to 1c, please refer to Figures 2a and 2b. Figure 2a illustrates an artificial intelligence voice-activated banking transaction system according to an embodiment of the present invention. Figure 2b is an operation flow chart of an artificial intelligence voice-activated bank transaction system executing instruction confirmation according to an embodiment of the present invention.

As shown in Figure 2a, the identity verification operation S40 performed by the computing device includes: Step S400: Retrieving a pending identity data contained in the first voice message through a speech recognition model; Step S401: Comparing the pending identity data with the registered identity data And confirm that the pending identity information matches the registered identity information, and step S402: when it is determined that the pending identity information matches the registered identity information, the computing device can perform an instruction confirmation operation on the user. For example, in step S400, the ATM may first send a message to the user: "Please tell the user account and password.", and the user then tells the voice receiving device the account number and password composed of English and numbers (i.e. First voice message), the voice receiving device transmits the acoustic signal to the memory for analysis by the computing device. In this case, the computing device can actively analyze the acoustic signal using a model including phonemes of English and numbers to obtain a string of input account numbers and input passwords (ie, pending identity information). Of course, in other embodiments, the user does not have to provide the account number and password, and it can also be the user's name, ID card number, birthday and other information. The implementation method is the same as above and this case is not limited thereto. For example, in step S401, the computing device compares the input account number and input password with a pre-stored registration account number and registration password (ie, registration identity information). Furthermore, the computing device will first compare the account to determine the corresponding password of the entered account, and then compare the entered password with the pre-stored registration password. If both are consistent, the identity verification operation is passed. Of course, this step can also compare other user information such as name, birthday, etc., which will not be described here.

As shown in Figure 2b, the instruction confirmation operation S50 performed by the computing device includes: Step S500: The computing device retrieves a pending instruction data contained in the second voice message through a speech recognition model; Step S501: Compares the pending instruction data and instructions. Multiple execution instructions in the database order data, step S502: determine an execution instruction data corresponding to the pending instruction data, and step S503: transmit the execution instruction data to the ATM to execute related services. For example, in step S500, the ATM can first send a message to the user: "Please describe the business you want to perform, including deposits, withdrawals, transfers...", where the business options can be included in In the command database 33 of Figure 1c, the user tells the voice receiving device the business content to be executed (ie, the second voice message), and the voice receiving device transmits the acoustic wave signal to the memory for analysis by the computing device. In this case, the computing device can actively analyze the acoustic signal using a model containing phonemes of relevant service content to obtain relevant service options (ie, pending instruction data). For example, in step S501, the computing device combines the pending instruction data (such as "transfer") and the execution instruction data (such as "withdrawal", "transfer", "deposit", etc.) pre-stored in the instruction database. Comparison is performed so that an execution instruction data (in this example, "transfer") can be determined in step S502, and the execution instruction data is sent to the ATM to perform related services.

Specifically, the user can describe the service item of "transfer", and the computing device can determine the execution instruction data as "transfer" and allow the ATM to provide the service. In other embodiments, the memory may further include a natural language processing model for allowing the computing device to consider usage in specific situations. For example, when the business item described by the user is "I want to get money", the computing device can first capture the keyword "money" through the natural language model and reduce the weight of the less relevant "I want" to further understand "Get money" is synonymous with the "withdraw" command in the command database to avoid the situation where "I want to get money" cannot be mapped to the "withdraw" command.

It should be noted that the execution of related services described in step S503 in this example can be understood as the execution of transfer-related services, including requesting the transfer from the user through the ATM. Account number and amount, etc., that is to say, the user will send another second voice message to the voice receiving device again, allowing the computing device to retrieve and analyze further information. In addition, step S503 is optionally set. For example, when the transaction is completed, the computing device may directly send the transaction record to the cloud server without sending the execution instruction data to the ATM.

Please refer to FIG. 3 , which is another operational flow chart of the artificial intelligence voice-activated banking transaction system performing identity verification according to an embodiment of the present invention. As shown in Figure 3, in the identity verification operation (S60), when the computing device retrieves a pending identity information contained in the first voice message through the speech recognition model (step S600), the computing device uses the automatic teller machine to The user sends an identity prompt message and receives an identity confirmation message from the user through the voice receiving device (step S601). For example, when the user reports that the account number is "abc123", the ATM can respond with the identity prompt message of "abc123", allowing the user to confirm whether the voice message received by the system is correct and respond with a "yes" or "no" (i.e. identity confirmation message). Then in step S602: the computing device determines whether the identity confirmation message is a positive message or a negative message. For example, when the ATM responds to the identity prompt message "abc723", the user can respond "no" (ie, a negative message), so that the system must re-perform the above identity verification operation to obtain another first voice message. The so-called re-executing the identity verification operation means that the user needs to re-narrate the first voice message containing the identity information. Specifically, when the computing device determines that the identity confirmation message is a negative message, it will send a request through the ATM to require the user to provide another first voice message to update the pending identity information, and then return to step S600. On the other hand, if the identity confirmation message is a positive message, the computing device maintains the pending identity data and executes step S603: the computing device compares the pending identity data and the registered identity data to determine that the pending identity data passes the identity verification operation, and step S604: Luck The computing device performs instruction confirmation operations on the user. These two steps are the same as the above-mentioned steps S401 and S402, and will not be described again here. Specifically, when the computing device determines that the pending identity information does not match the registered identity information and determines that the pending identity information has not passed identity verification, the computing device may wait for a new first voice message to be input or require the user to enter a new third voice message through the ATM. A voice message.

Please refer to FIG. 4 , which is another operation flow chart of the artificial intelligence voice-activated banking transaction system performing operation confirmation according to an embodiment of the present invention. As shown in Figure 4, in the instruction confirmation operation (S70), when the computing device determines an execution instruction data corresponding to the pending instruction data (step S700), the computing device can send an instruction prompt message to the user through the ATM, And by receiving a command confirmation message from the user from the voice receiving device (step S701). For example, when the user says "I want to withdraw two thousand dollars", the ATM can send out a command prompt message "Do you want to withdraw money?", and the user then replies with a "Yes" command confirmation message. Regarding how to correspond from "receiving two thousand yuan" to "withdrawing money", you can refer to the description of the direct conversation model or natural language model mentioned above, and will not be described in detail here. Next, in step S702, the computing device determines whether the instruction confirmation message is a positive message or a negative message. If it is a positive message, the computing device can send the execution instruction data to the ATM to perform related services (step S703). Alternatively, when the instruction confirmation message is a negative message, the instruction confirmation operation can be re-executed to obtain another execution instruction data. That is to say, when the computing device determines that the instruction confirmation message is a non-agreed message, the computing device can send a request through the ATM to require the user to provide another second voice message to update the pending instruction data. For example: "Do you want to withdraw money?" (instruction prompt message), "No" (instruction confirmation message), "Excuse me What kind of business is going on? "(Request the second voice message), "Transfer" (update the pending instruction data), "Do you want to transfer money?" (Instruction prompt message), "Yes" (Instruction confirmation message).

The identity verification of the modified embodiment of the system in this case can further include the image recognition function. Please refer to Figures 5a to 5c. Figure 5a is a block diagram of an artificial intelligence voice-controlled banking transaction system according to another embodiment of the present invention. Figure 5b is a block diagram of an artificial intelligence voice-activated banking system according to another embodiment of the present invention. A block diagram of the usage scenario of the transaction system. Figure 5c is a block diagram of the memory of the artificial intelligence voice-activated bank transaction system receiving voice messages and facial image data according to another embodiment of the present invention.

As shown in Figures 5a to 5c, the artificial intelligence voice-activated banking transaction system 1' not only includes an automatic teller machine 10, a computing device 20, a memory 30', and a voice receiving device 40, but also includes an image capture device 50 connected to the memory. 30'. As shown in FIG. 5b , the image capture device 50 may be, for example, a camera installed on the ATM 10 for capturing facial image data of the user C. As shown in Figure 5c, the memory 30' further stores a facial feature detection model 34 and a registered facial feature 35 of the user in advance.

Please refer to FIG. 6 , which is an operation flow chart of an artificial intelligence voice-activated banking transaction system performing facial recognition for identity verification according to another embodiment of the present invention. As shown in Figure 6, the identity verification operation performed by the computing device further includes a facial recognition operation, including step S900: capturing a pending facial feature contained in the facial image data through the facial feature detection model, Step S901: Compare the pending facial features with the logged-in facial features. Step S902: Determine whether the pending facial features match the logged-in facial features. If the pending facial features match the logged-in facial features, proceed to steps S903: The computing device performs additional authentication The operation or the instruction confirms the operation, and if the pending facial features do not match the logged-in facial features, the computing device re-executes the identity verification operation.

For example, after taking an image of the user's face, the image capture device can transfer the image data to the memory, and the computing device can capture the characteristics of the image data through a facial feature detection model, where the face The partial feature detection model can be a neural network with deep learning capabilities (such as VGG-Face, Google FaceNet, OpenFace, etc.). Taking Google FaceNet as an example, it does not output classification results at the beginning, but first outputs quantified feature values. This operation can extract facial features such as contours into a to-be-determined facial feature, and then use it in the feature vector space. Different facial photos are compared to determine the similarity of different faces. In other words, the closer the two points are in the feature vector space, the higher the similarity of the faces. In addition, the login facial features stored in the memory are associated with the above-mentioned registered identity information. Specifically, when the user first performs the comparison of identity information as shown in step S401 in Figure 2a, the identity information will correspond to a login face. features for comparison. In terms of comparison and judgment of facial features, the recognition accuracy can be increased by capturing multiple images of the user, which will not be described in detail here. It should be noted that when the computing device determines that the undetermined facial features do not match the logged-in facial features, it means that the user's identity information may be incorrect and the identity verification operation needs to be performed again. Therefore, the above-mentioned facial recognition operation will be performed again. However, Figure 6 does not show all the identity verification operations, so the negative arrow in step S902 is shown pointing to step S900. This case is not limited thereto.

Please refer to FIG. 7 . FIG. 7 is a block diagram illustrating another memory of the artificial intelligence voice-activated banking transaction system receiving messages and data according to another embodiment of the present invention. As shown in Figure 7, the memory 30" further stores a lip recognition model 36 (such as LipNet) in advance, and the computing device is further used to capture multiple lip images of the user through the image capture device 50 to obtain A dynamic lip shape data is obtained, and the instruction confirmation operation performed by the computing device 20 further includes a lip shape recognition operation.

Please refer to FIG. 8 , which is an operation flow chart of an artificial intelligence voice-activated banking transaction system executing lip recognition for instruction confirmation according to yet another embodiment of the present invention. As shown in Figure 8, the lip recognition operation S110 performed by the computing device includes step S1100: acquiring a reference command data included in the dynamic lip data through the lip recognition model, step S1101: judging the pending command data and the reference command data. Whether they are the same, and step S1102: retain the pending instruction data. In step S1101, if it is determined that the pending instruction data is the same as the reference instruction data, then proceed to step S1102. If it is determined that the pending instruction data is different from the reference instruction data, it means that the pending instruction data obtained by the speech recognition model and the reference instruction obtained by the lip recognition model If the data is different, the above command confirmation operation can be performed again. It should be noted that this figure does not show all the command confirmation operations, so the negative arrow indicating "re-perform the command confirmation operation" points to step S1100 (meaning that when the command confirmation operation is re-performed, the lip recognition operation can also be re-performed) , this case is not limited to this. Specifically, the principle of the lip shape recognition model is similar to the above-mentioned speech recognition model. It can decompose spoken language into mouth shape elements corresponding to different pronunciations, and store them as a sequence for analysis by the computing device, which will not be described in detail here. It should be noted that the image capture device used in the foregoing embodiments may only have the function of capturing images, but in this embodiment, the image capture device needs to have the function of recording continuous videos or capturing high-speed continuous images to achieve lip shape. Identify.

Please refer to FIG. 9 , which is a block diagram of a usage scenario of an artificial intelligence voice-activated banking transaction system according to other embodiments of the present invention. As shown in FIG. 9 , the voice receiving device 40 can be a mobile device (such as a smart phone) of user C, and the voice receiving device 40 can be changed. Connected to the computing device 20. In this example, user C's mobile phone can be connected to the computing device 20 and the memory 30. Specifically, the mobile phone can be connected to the ATM 10 and then connected to the computing device 20 and the memory 30. In this case Not limited to this. User C can tell his personal information and the banking business he wants to perform to his mobile phone, making it difficult for others around him to hear the private information. In the same way, the system can send a prompt message to user C through the mobile phone without being heard by the people around him. In addition, the mobile phone can also be used to receive an encrypted prompt message from the system and convert the prompt message into a prompt tone and then play it to the user, thereby increasing information security.

Please refer to FIG. 10 , which is a schematic block diagram of other memories of the artificial intelligence voice-activated banking transaction system receiving messages and data according to other embodiments of the present invention. As shown in FIG. 10 , the memory 30'' further stores a first trigger message 37 and a second trigger message 38 in advance, and is further used to receive a third voice message from the voice receiving device.

Please refer to FIG. 11 in conjunction with FIG. 10 . FIG. 11 is a block flow chart of an artificial intelligence voice-activated banking transaction system when triggered according to other embodiments of the present invention. As shown in Figure 11, when the computing device receives a third voice message from the user through the voice receiving device, it can determine whether the third voice message matches the first trigger message or the second trigger message. If the third voice message matches the first trigger message, Once the message is triggered, the above identity verification operation is re-executed. If the third voice message matches the second trigger message, the command confirmation operation is re-executed. Otherwise, no additional operation is performed. Specifically, if a user completes the identity verification message and finds that he/she wants to operate another account, he/she can send a third voice message that matches the first trigger message to the voice receiving device (such as "re-verification"). ”). Similarly, if a user says that he/she wants to transfer money and then finds that he/she actually wants to withdraw money, he/she can send a third voice message (such as "re-transaction") that matches the second trigger message to the voice receiving device. Of course, as long as you use If the person avoids talking about the first trigger message or the second trigger message, the system will not be triggered by the third voice message to interrupt the transaction.

Based on the disclosure of this article, a user scenario described below can be achieved. It should be noted that this scenario is only an example, and this case is not limited to this. After the customer goes to the ATM and follows the instructions to insert a debit card or click on the card-not-present transaction function, the ATM plays a voice instructing the customer to approach the microphone within a specified time and tell them identity verification questions, such as birthday, phone number, account number, password and other personal information. The microphone receives the customer's voice, filters the noise through noise reduction technology, and transmits it to the system in this case to determine the content of the customer's voice. The system determines that the identity verification has passed, and the voice played in the ATM instructs the customer to approach the microphone within the specified time and describe the transaction items (deposit/withdrawal/transfer) and transaction content (account number/amount). If the system determines that the identity verification has failed or exceeded , go back to the identity verification steps above. The microphone receives the customer's voice, the system identifies the customer's transaction content, and plays the voice again on the ATM to instruct the customer to approach the microphone within the specified time and reply to the transaction confirmation/modification/cancellation instructions. The system receives the customer's confirmation instruction and executes the transaction at the ATM. If the system receives the customer's modification instruction, it returns to the above corresponding process. If the system receives the customer's cancellation or the system transaction times out, it returns to the initial process. After the customer completes the transaction, the system sends the retained receipt via email and SMS.

With the above structure, the artificial intelligence voice-controlled banking transaction system disclosed in this case can perform content judgment on the received voice messages through a memory and computing device that stores an artificial intelligence recognition model, so as to safely and effectively perform identity verification and command control. function. Users only need to tell their identity information in front of the ATM to complete the verification, and then verbally express the banking services they want to perform. In this way, general banking services can be completed completely through voice control, which is suitable for various ethnic groups. They are all convenient measures.

Although the present invention is disclosed in the foregoing embodiments, they are not intended to limit the present invention. All changes and modifications made without departing from the spirit and scope of the present invention shall fall within the scope of patent protection of the present invention. Regarding the protection scope defined by the present invention, please refer to the attached patent application scope.

1: Artificial intelligence voice-controlled banking transaction system

10:ATM

20:Computing device

30:Memory

40: Voice receiving device

Claims (8)

An artificial intelligence voice-activated bank transaction system includes: an automatic teller machine for allowing a user to handle bank counter services by themselves; a computing device connected to the automatic teller machine for performing an identity verification operation and an instruction on the user Confirm operation; a memory connected to the computing device for storing a first voice message and a second voice message, and pre-stored a voice recognition model, a registered identity information of the user and a command database; A voice receiving device connected to the memory for receiving the first voice message and the second voice message of the user and transmitting them to the memory, wherein the voice receiving device is an earphone device; and a mobile device , connected to the ATM and the computing device, for receiving an encrypted prompt message from the computing device and converting the prompt message into a prompt sound and then playing it to the user, wherein the identity verification operation performed by the computing device including: capturing a pending identity data contained in the first voice message through the speech recognition model; and comparing the pending identity data with the registered identity data and confirming that the pending identity data matches the registered identity data, and the calculation The command confirmation operation performed by the device includes: retrieving a pending command data contained in the second voice message through the voice recognition model; and Compare the pending instruction data with multiple execution instruction data in the instruction database to determine an execution instruction data corresponding to the pending instruction data. The artificial intelligence voice-activated banking transaction system as described in claim 1, wherein the identity verification operation performed by the computing device further includes: when the computing device retrieves the pending identity contained in the first voice message through the speech recognition model When receiving data, execute: send an identity reminder message to the user through the ATM; the computing device receives an identity confirmation message from the user from the voice receiving device, and determines that the identity confirmation message is a positive message or a negative message; if the identity confirmation message is a positive message, the computing device maintains the pending identity information; and if the identity confirmation message is a negative message, the computing device issues a request through the ATM to require the user to provide another A first voice message to update the pending identity information. The artificial intelligence voice-activated banking transaction system as described in claim 1, wherein the instruction confirmation operation performed by the computing device further includes: when the computing device determines an execution instruction data corresponding to the pending instruction data, execute: through the automatic teller machine Send a command prompt message to the user; receive a command confirmation message from the user from the voice receiving device, and determine whether the command confirmation message is an consent message or a non-agree message; If the instruction confirmation message is the consent message, the computing device sends the execution instruction data to the ATM to perform related services; if the instruction confirmation message is the non-agree message, the computing device sends a request through the ATM to request the The user provides another second voice message to update the pending instruction data. The artificial intelligence voice-activated banking transaction system as described in claim 1, wherein the instruction confirmation operation performed by the computing device further includes: when the computing device determines an execution instruction data corresponding to the pending instruction data, execute: convert the execution instruction The information is transmitted to the ATM to perform related services. The artificial intelligence voice-activated banking transaction system as described in claim 1 further includes an image capture device connected to the memory for capturing facial image data of the user, and the memory further stores a facial image data in advance. A facial feature detection model and a logged-in facial feature of the user, and the logged-in facial feature is associated with the registered identity information, wherein the identity verification operation performed by the computing device further includes a facial recognition operation, including : Acquire an undetermined facial feature contained in the facial image data through the facial feature detection model; compare the undetermined facial feature with the registered facial feature to determine whether the undetermined facial feature matches the Log in facial features; if the pending facial features match the logged in facial features, the computing device performs other identity verification operations or the command confirmation operation; If the pending facial feature does not match the registered facial feature, the computing device sends a request through the ATM to require the user to provide another first voice message to update the pending identity information and the registered identity information. The artificial intelligence voice-activated banking transaction system as described in claim 5, wherein the memory further stores a lip recognition model, and the computing device is further used to capture multiple lip images of the user through the image capture device. Obtaining a dynamic lip shape data, and the command confirmation operation executed by the computing device further includes a lip shape recognition operation, including: retrieving a reference command data included in the dynamic lip shape data through the lip shape recognition model; judging Whether the pending order data is the same as the reference order data; if the pending order data is the same as the reference order data, the pending order data is retained; and if the pending order data is different from the reference order data, the computing device uses the ATM A request is issued to require the user to provide another second voice message to update the pending instruction data. The artificial intelligence voice-activated banking transaction system as described in claim 1, wherein the memory further stores a first trigger message and a second trigger message, so that when the computing device receives a first trigger message from the user through the voice receiving device, When three voice messages are sent, it is determined whether the third voice message matches the first trigger message or the second trigger message. If the third voice message matches the first trigger message, the identity verification operation is re-executed. If the third voice message matches the first trigger message, the identity verification operation is re-executed. If the voice message matches the second trigger message, the command confirmation operation is re-executed, otherwise, no additional operation is performed. The artificial intelligence voice-activated banking transaction system according to claim 1, wherein the voice receiving device has a filter for filtering out noise in the first voice message and the second voice message.

Images