AI-based intelligent voice outbound calling system

Through the intelligent voice outgoing call system based on AI, the use of semantic understanding and in-depth analysis models to determine the outgoing call range and promote voice, the problem of low voice outgoing call efficiency in the existing technology is solved, and a higher business promotion success rate and user Satisfaction.

CN118612343BActive Publication Date: 2024-11-26HANGZHOU XUANQIN INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411092842.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-11-26
Estimated Expiration
2044-08-09

AI Technical Summary

Technical Problem

The existing voice outbound call system has low service efficiency and is easily hung up due to dissatisfaction with telecom users. How to improve the efficiency of voice outbound call is a technical problem that needs to be solved urgently.

Method used

Adopt an intelligent voice outbound call system based on AI, determine the business topic type through service text data, accurately determine the outbound call range, and select appropriate service promotion voice based on the identity information of telecom users after being connected, and use semantic understanding model and in-depth analysis model. Optimize outbound call strategies and streamline or adjust promotion content to increase the duration of connection maintenance.

Benefits of technology

It improves the success rate and efficiency of business promotion. By automatically analyzing business customer data and voice characteristics, accurately match user needs, reduce hangups, and improves user experience and promotion effects.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118612343B_ABST
    Figure CN118612343B_ABST
Patent Text Reader

Abstract

The present invention belongs to the technical field of voice outbound calls. An AI-based intelligent voice outbound call system is provided, the system comprising an outbound call range determination module, an outbound call module and an outbound call control module; the input end of the outbound call control module is connected to the outbound call range determination module, and the output end of the outbound call control module is connected to the outbound call module. The solution of the present invention can realize automatic analysis of the business text data of the business customer to determine the business subject type involved in this outbound call, thereby accurately determining the appropriate outbound call range, and can select appropriate business promotion voice based on the identity information of the connected telecommunications user, which can improve the success rate of business promotion.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of voice outbound calling technology, and in particular to an AI-based intelligent voice outbound calling system. Background Art

[0002] Voice outbound calling is a technology that automatically dials a phone through a telephone system and plays pre-recorded or real-time generated voice information. It is widely used in customer service, marketing, reminder services and other fields. The existing voice outbound calling system basically calls telecom users within a specified range, and then promotes services based on fixed speech segments after the call is connected. However, the efficiency of business promotion in this way is low, and it is easy for telecom users to hang up due to dissatisfaction. How to improve the efficiency of voice outbound calling is a technical problem that needs to be solved urgently. Summary of the invention

[0003] In this regard, the present invention provides an AI-based intelligent voice outbound calling system, electronic device, computer storage medium and computer program product to solve at least one of the above-mentioned technical problems.

[0004] The present invention provides an AI-based intelligent voice outbound call system, the system comprising an outbound call range determination module, an outbound call module and an outbound call control module; the input end of the outbound call control module is connected to the outbound call range determination module, and the output end of the outbound call control module is connected to the outbound call module;

[0005] The outbound call range determination module is used to determine the service subject type according to the service text data, and determine the outbound call range according to the service subject type; wherein the outbound call range includes a geographical range and a number range;

[0006] The outbound call control module is used to generate a first control instruction according to the outbound call range;

[0007] The outbound call module is used to automatically initiate a call to each number within the outbound call range based on the first control instruction, and feed back the other party's voice after the call is connected to the outbound call control module;

[0008] Wherein, the outbound call control module is also used for:

[0009] The identity information of the connected party is determined according to the other party's voice, a service promotion voice is determined according to the identity information, and a second control instruction is generated according to the service promotion voice; wherein the second control instruction is used to control the outbound call module to broadcast the service promotion voice to the connected party.

[0010] Optionally, determining the service subject type according to the service text data, and determining the outbound call range according to the service subject type includes:

[0011] Using a semantic understanding model to perform semantic understanding on the business text data to obtain semantic features, and matching and analyzing the semantic features with standard semantic features corresponding to each known business subject type in the database;

[0012] If the match is successful, the standard outbound call range corresponding to the known business subject type that is successfully matched is determined as the outbound call range;

[0013] If the match fails, the semantic feature, the standard semantic features corresponding to several known business subject types with high matching degree, and the standard outbound call range are input into the deep analysis model, and the deep analysis model outputs the outbound call range.

[0014] Optionally, the semantic understanding model is built based on BERT, and the deep analysis model is built based on ChatGPT.

[0015] Optionally, determining identity information of the connected other party according to the other party's voice, and determining the service promotion voice according to the identity information includes:

[0016] Extracting voiceprint features from the other party's voice within a preset time period, and if the other party's voice does not contain the specified vocabulary, determining the first identity information of the other party according to the extracted voiceprint features;

[0017] If the other party's voice contains a designated word, determining whether the designated word matches the identity information of the other party's owner, and if so, using the identity information of the owner as the second identity information;

[0018] The service promotion voice is determined according to the first identity information or the second identity information.

[0019] Optionally, determining the service promotion voice according to the identity information further includes:

[0020] monitoring the duration of the connection of the first party after the call is connected, and if the duration of the connection is less than a duration threshold, determining that the voice outbound call to the first party has failed;

[0021] Simplifying the service promotion voice corresponding to the first counterparty to obtain a simplified service promotion voice;

[0022] After the second party is successfully connected, controlling the outbound call module to broadcast the simplified service promotion voice to the connected second party;

[0023] The first party and the second party are geographical neighbors.

[0024] Optionally, the performing simplified processing on the service promotion voice corresponding to the first counterparty includes:

[0025] The small talk in the business promotion voice is removed, and / or greater preferential information is added to the business promotion voice.

[0026] Optionally, determining the service promotion voice according to the identity information further includes:

[0027] Identify third and fourth parties who are close neighbors and close relatives;

[0028] Determine, according to the idle states of the handheld devices of the third party and the fourth party, any one of them as a first-call object, and initiate a voice outbound call to the first-call object;

[0029] If the connection maintenance duration is longer than the duration threshold, the voice outbound call to the first-called party is deemed successful, and no voice outbound call is initiated to the third party or the fourth party other than the first-called party.

[0030] The present invention also provides an electronic device, which is applied to an AI-based intelligent voice outbound call system as described in any of the above items, and the electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor.

[0031] The present invention also provides a computer storage medium, which is applied to an AI-based intelligent voice outbound call system as described in any of the above items, and the computer storage medium stores a computer program that can be executed by a processor.

[0032] The present invention also provides a computer program product, which is applied to an AI-based intelligent voice outbound call system as described in any of the above items, and the computer program product includes a computer program that can be executed by a processor.

[0033] The solution of the present invention can realize automatic analysis of the business text data of the business customer to determine the business subject type involved in the outbound call, thereby accurately determining the appropriate outbound call range, and can select the appropriate business promotion voice based on the identity information of the connected telecommunications user, thereby improving the success rate of business promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1It is a structural schematic diagram of an AI-based intelligent voice outbound calling system disclosed in an embodiment of the present invention;

[0036] Figure 2 It is a schematic diagram of a process for determining an outbound call range according to a service subject type disclosed in an embodiment of the present invention;

[0037] Figure 3 It is a structural schematic diagram of an electronic device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following is a description of the implementation of the present application by specific specific embodiments. People familiar with the technology can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0039] In addition, the technical features involved in the different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0040] See also Figure 1 , an embodiment of the present invention discloses an AI-based intelligent voice outbound call system, the system comprising an outbound call range determination module, an outbound call module and an outbound call control module; the input end of the outbound call control module is connected to the outbound call range determination module, and the output end of the outbound call control module is connected to the outbound call module;

[0041] The outbound call range determination module is used to determine the service subject type according to the service text data, and determine the outbound call range according to the service subject type; wherein the outbound call range includes a geographical range and a number range;

[0042] The outbound call control module is used to generate a first control instruction according to the outbound call range;

[0043] The outbound call module is used to automatically initiate a call to each number within the outbound call range based on the first control instruction, and feed back the other party's voice after the call is connected to the outbound call control module;

[0044] Wherein, the outbound call control module is also used for:

[0045] The identity information of the connected party is determined according to the other party's voice, a service promotion voice is determined according to the identity information, and a second control instruction is generated according to the service promotion voice; wherein the second control instruction is used to control the outbound call module to broadcast the service promotion voice to the connected party.

[0046] The present invention first determines the business theme type of the promotion business involved in this voice outbound call based on the business text data that needs to be promoted, such as telecommunications packages (traffic, roaming, family numbers, etc.), banking services, fund services, etc., and accordingly determines the best outbound call range, including the geographical range and number range. Among them, the geographical range can be the administrative division of a prefecture-level city, a radius of 100 kilometers, etc., and the number range can be the 138 segment, the 158 segment, or a group of specific telecommunications numbers. The telecommunications users corresponding to the group of telecommunications numbers can be telecommunications users corresponding to the above-mentioned business theme types, such as stock speculators, car enthusiasts, etc., which can be determined through big data mining technology.

[0047] The outbound call control module can control the outbound call module to call each telecommunications number within the range according to the above-mentioned outbound call range. After the call is connected, the outbound call control module can receive feedback from the outbound call module about the voice of the connected party, based on which its identity information can be determined, thereby determining the best business promotion voice, and starting business promotion communication with the user accordingly.

[0048] Therefore, the solution of the present invention can realize automatic analysis of the business text data of the business customer to determine the business subject type involved in the outbound call, thereby accurately determining the appropriate outbound call range, and can select the appropriate business promotion voice based on the identity information of the connected telecommunications user, thereby improving the success rate of business promotion.

[0049] The business text data is generated by the business customer, and may be a detailed introduction booklet of the business to be promoted, or a brief introduction content, or a poster page of the business to be promoted, etc., and the present invention is not limited thereto.

[0050] Optionally, refer to Figure 2 As shown, the determining of the service subject type according to the service text data, and the determining of the outbound call range according to the service subject type include:

[0051] Using a semantic understanding model to perform semantic understanding on the business text data to obtain semantic features, and matching and analyzing the semantic features with standard semantic features corresponding to each known business subject type in the database;

[0052] If the match is successful, the standard outbound call range corresponding to the known business subject type that is successfully matched is determined as the outbound call range;

[0053] If the match fails, the semantic feature, the standard semantic features corresponding to several known business subject types with high matching degree, and the standard outbound call range are input into the deep analysis model, and the deep analysis model outputs the outbound call range.

[0054] In this embodiment, the business customer may not provide the detailed content of the business to be promoted, but may only provide a simple text such as a poster page of the business to be promoted, and the semantic understanding model may be used to perform semantic understanding on it, and the corresponding semantic features may be obtained. If the semantic feature successfully matches each known business theme type in the database, it means that it is a known business type, such as a telecommunications package, banking business, and fund business. At this time, the standard outbound call range corresponding to the successfully matched known business theme type may be directly determined as the outbound call range. If the match fails, it means that it is a new business type that has not been recorded or implemented by the outbound call system, such as volunteer recruitment. At this time, the semantic features, the standard semantic features corresponding to several (for example, the top three) known business theme types with the highest matching degree, and the standard outbound call range may be input into the deep analysis model, and the deep analysis model outputs the most suitable outbound call range through deep analysis, that is, determines which types of numbers in what geographical range to initiate outbound calls.

[0055] Optionally, the semantic understanding model is built based on BERT, and the deep analysis model is built based on ChatGPT.

[0056] In this embodiment, the semantic understanding model in the present invention may be based on BERT, which is called Bidirectional Encoder Representations from Transformers, a language model training method using massive texts, which is widely used in a variety of natural language processing tasks, such as text classification, text matching, machine reading comprehension, etc. The deep analysis model may be based on ChatGPT, which is an artificial intelligence technology-driven natural language processing tool launched by OpenAI. It uses the Transformer neural network architecture, which can not only conduct conventional language conversations, but also obtain a powerful dedicated model through fine-tuning. The present invention uses training data marked with outbound call ranges to fine-tune and train it to obtain the deep analysis model. The format of the training data may be [semantic features, standard semantic features corresponding to known business topic type n, standard outbound call range corresponding to known business topic type n, label: outbound call range].

[0057] Optionally, determining identity information of the connected other party according to the other party's voice, and determining the service promotion voice according to the identity information includes:

[0058] Extracting voiceprint features from the other party's voice within a preset time period, and if the other party's voice does not contain the specified vocabulary, determining the first identity information of the other party according to the extracted voiceprint features;

[0059] If the other party's voice contains a designated word, determining whether the designated word matches the identity information of the other party's owner, and if so, using the identity information of the owner as the second identity information;

[0060] The service promotion voice is determined according to the first identity information or the second identity information.

[0061] In this embodiment, the voiceprint feature extraction is performed on the voice of the other party within the first preset time (for example, 2s) after the call is connected, and the identity information of the other party who is connected can be determined based on the voiceprint feature. Among them, the user of a terminal with a telecommunications number is generally one person (that is, the owner), and in some cases it can also be multiple people, such as parents using their ID cards to open a telecommunications number for their children. Therefore, a correspondence between each voiceprint feature and the preset identity information can be established in advance, that is, the voiceprint features (children and parents' voiceprints) are collected when the telecommunications number is opened. Of course, voiceprint features can also be collected in advance by other means, such as obtaining microphone permissions at the mobile phone APP. By matching the voiceprint features with the voiceprint features corresponding to each preset identity information in the corresponding relationship, the identity information of the other party who is connected can be filtered out from the associated multiple identity information.

[0062] At the same time, the receiver of the terminal of the telecommunications number may not be the owner of the phone in some cases, but may be the owner's family, colleagues, etc. In this regard, the present invention is set to simultaneously analyze the designated words during the voiceprint extraction process. If the designated words are detected, it means that the receiver is not the owner of the phone, otherwise it means that the receiver is the owner of the phone. Among them, the designated words are, for example, "Mom", "Zhang Xiaolan (owner's name)", "Wife", etc. When the receiver says the above designated words, it means that the receiver is most likely not the owner of the phone, but temporarily answers the call and transfers it to the owner. The designated words correspond to the first identity information of the owner. When the owner is a married woman, the designated words are the above content, and when the owner is a married man, the designated words can be "Dad", "Wang Dali (owner's name)", "Husband", etc., and the details will not be repeated. The receiver may say a common designated word for some reason, but the designated word may not match the identity of the owner, and no response is given at this time. For example, the designated word detected is "Dad", but the owner is a woman, and no response is given at this time, that is, the receiver is still considered to be the owner. When it is detected that the other party's voice contains a designated word and the designated word matches the identity information of the other party's owner, the owner's identity information corresponding to the telecommunications number is used as the second identity information.

[0063] After the identity of the other party is determined, a personalized business promotion voice can be determined accordingly, thereby improving the success rate of business promotion. For example, when the identity information is the owner of the phone, since the owner's preferences can be extracted through big data analysis technology, a corresponding business promotion voice adapted to the preferences can be generated. For example, big data analysis shows that the owner is prone to hang up the phone after saying "Hello, sir / madam..." after the call is connected (identified as a sales call), resulting in no chance to broadcast the subsequent business content. At this time, the subsequent business content can be broadcast directly, that is, the above-mentioned greeting part is cancelled. When the identity information is not the owner of the phone, a standard business promotion voice can be broadcast.

[0064] It should be noted that the second identity information is the owner's identity, while the first identity information can be the owner's identity, other preset identities, or unidentified identities. Unidentified identities refer to the current voiceprint features not matching any pre-associated identity information, and the standard service promotion voice is also broadcasted at this time.

[0065] Optionally, determining the service promotion voice according to the identity information further includes:

[0066] monitoring the duration of the connection of the first party after the call is connected, and if the duration of the connection is less than a duration threshold, determining that the voice outbound call to the first party has failed;

[0067] Simplifying the service promotion voice corresponding to the first counterparty to obtain a simplified service promotion voice;

[0068] After the second party is successfully connected, controlling the outbound call module to broadcast the simplified service promotion voice to the connected second party;

[0069] The first party and the second party are geographical neighbors.

[0070] In this embodiment, voice outbound calls are usually made in batches, so that two or more people in the same geographical area will receive outbound calls one after another. When the duration of the connection with the first party is lower than the duration threshold, it means that the adaptability of the broadcasted business promotion voice is not enough, which may be that the business type is not suitable, or the business promotion voice contains too many greetings, resulting in the first party not being interested in listening to it, resulting in hanging up. In this regard, when the present invention identifies a second party in the same geographical proximity as the first party, it broadcasts a streamlined business promotion voice to the second party, which can increase the duration of the connection with the second party, thereby increasing the probability of successful business promotion.

[0071] It should be noted that the first party and the second party who are close neighbors can be determined by analyzing their respective geographical location information, and the geographical location information can be obtained from a mobile phone APP terminal, which will not be described in detail.

[0072] Optionally, the performing simplified processing on the service promotion voice corresponding to the first counterparty includes:

[0073] The small talk in the business promotion voice is removed, and / or greater preferential information is added to the business promotion voice.

[0074] In this embodiment, the simplified service promotion voice may be a service promotion voice without greetings, or a service promotion voice that provides greater discounts (preferably without greetings). The above methods can improve the success rate of voice outbound calls.

[0075] Optionally, determining the service promotion voice according to the identity information further includes:

[0076] Identify third and fourth parties who are close neighbors and close relatives;

[0077] Determine, according to the idle states of the handheld devices of the third party and the fourth party, any one of them as a first-call object, and initiate a voice outbound call to the first-call object;

[0078] If the connection maintenance duration is longer than the duration threshold, the voice outbound call to the first-called party is deemed successful, and no voice outbound call is initiated to the third party or the fourth party other than the first-called party.

[0079] In this embodiment, the third party and the fourth party of the neighbors may also be close relatives. For example, by analyzing historical positioning data, it is found that the two often travel together, or the handheld terminals of the two are often interconnected (such as Bluetooth connection), and they participate in game tasks in the same online game hall, etc. People with close relationships are more willing to share the content of incoming calls, especially business promotion calls. Therefore, the present invention is configured to initiate a voice outbound call to one of the third and fourth parties whose handheld devices are in an idle state. If the connection maintenance duration of the first call object is higher than the duration threshold, it means that the voice outbound call to it is successful. At this time, there is no need to initiate outbound calls to the remaining other close neighbors, thereby improving the efficiency of outbound calls and reducing harassment to users.

[0080] The idle state of the handheld device may be that the phone is in a screen-off state, the current interface is a main control interface, and the like.

[0081] See also Figure 3 The embodiment of the present invention also discloses that the present invention also provides an electronic device, which is applied to an AI-based intelligent voice outbound call system as described in any of the above items, and the electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor.

[0082] The present invention also provides a computer storage medium, which is applied to an AI-based intelligent voice outbound call system as described in any of the above items, and the computer storage medium stores a computer program that can be executed by a processor.

[0083] The present invention also provides a computer program product, which is applied to an AI-based intelligent voice outbound call system as described in any of the above items, and the computer program product includes a computer program that can be executed by a processor.

[0084] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0085] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An AI-based intelligent voice outbound calling system. Features: The system comprises an outbound call range determination module, an outbound call module and an outbound call control module; the input end of the outbound call control module is connected to the outbound call range determination module, and the output end of the outbound call control module is connected to the outbound call module; The outbound call range determination module is used to determine the service subject type according to the service text data, and determine the outbound call range according to the service subject type; wherein the outbound call range includes a geographical range and a number range; The outbound call control module is used to generate a first control instruction according to the outbound call range; The outbound call module is used to automatically initiate a call to each number within the outbound call range based on the first control instruction, and feed back the other party's voice after the call is connected to the outbound call control module; Wherein, the outbound call control module is also used for: Determine the identity information of the connected party according to the other party's voice, determine the service promotion voice according to the identity information, and generate a second control instruction according to the service promotion voice; wherein the second control instruction is used to control the outbound call module to broadcast the service promotion voice to the connected party; Determining identity information of the connected party according to the other party's voice, and determining a service promotion voice according to the identity information, including: Extracting voiceprint features from the other party's voice within a preset time period, and if the other party's voice does not contain the specified vocabulary, determining the first identity information of the other party according to the extracted voiceprint features; If the other party's voice contains a designated word, determining whether the designated word matches the identity information of the other party's owner, and if so, using the identity information of the owner as the second identity information; Determine a service promotion voice according to the first identity information or the second identity information; Determining the service promotion voice according to the identity information also includes: monitoring the duration of the connection of the first party after the call is connected, and if the duration of the connection is less than a duration threshold, determining that the voice outbound call to the first party has failed; Simplifying the service promotion voice corresponding to the first counterparty to obtain a simplified service promotion voice; After the second party is successfully connected, controlling the outbound call module to broadcast the simplified service promotion voice to the connected second party; wherein the first party and the second party are geographically close neighbors; Determining the service promotion voice according to the identity information also includes: Identify third and fourth parties who are close neighbors and close relatives; Determine, according to the idle states of the handheld devices of the third party and the fourth party, any one of them as a first-call object, and initiate a voice outbound call to the first-call object; If the connection maintenance duration is longer than the duration threshold, the voice outbound call to the first-called party is deemed successful, and no voice outbound call is initiated to the third party or the fourth party other than the first-called party.

2. According to claim 1, an AI-based intelligent voice outbound calling system, Features: Determining a service subject type according to the service text data, and determining an outbound call scope according to the service subject type, including: Using a semantic understanding model to perform semantic understanding on the business text data to obtain semantic features, and matching and analyzing the semantic features with standard semantic features corresponding to each known business subject type in the database; If the match is successful, the standard outbound call range corresponding to the known business subject type that is successfully matched is determined as the outbound call range; If the match fails, the semantic feature, the standard semantic feature corresponding to several known business subject types with high matching degree, and the standard outbound call range are input into the deep analysis model, and the deep analysis model outputs the outbound call range.

3. According to the AI-based intelligent voice outbound calling system of claim 2, Features: The semantic understanding model is built based on BERT, and the deep analysis model is built based on ChatGPT.

4. According to claim 1, an AI-based intelligent voice outbound calling system, Features: The streamlining of the service promotion voice corresponding to the first counterparty includes: The small talk in the business promotion voice is removed, and / or greater preferential information is added to the business promotion voice.

5. An electronic device, applied to an AI-based intelligent voice outbound calling system as claimed in any one of claims 1 to 4, Features: The electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor.

6. A computer storage medium, applied to an AI-based intelligent voice outbound calling system as claimed in any one of claims 1 to 4, Features: The computer storage medium stores a computer program that can be executed by a processor.

7. A computer program product, applied to an AI-based intelligent voice outbound calling system as claimed in any one of claims 1 to 4, Features: The computer program product comprises a computer program executable by a processor.