Intelligent outbound answering broadcast control method and device
By acquiring and segmenting the structured data of intelligent outbound call nodes, and constructing a broadcast data packet sequence, the problem of discontinuity and inconsistency in response content during intelligent outbound calls is solved, and ordered streaming broadcast is realized, improving the continuity and consistency of broadcast.
Patent Information
- Application Number
- CN202611452889.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-18
- Publication Date
- 2026-11-10
AI Technical Summary
In intelligent outbound calling, existing technologies suffer from issues such as repetitive response content, chaotic information from multiple sources, and inconsistent content after model stream interruption, resulting in discontinuous and inconsistent broadcasting.
By acquiring the structured data of the current outbound call node, the response broadcast is determined and segmented according to preset rules to construct a broadcast data packet sequence, including start, segmentation, and end data packets, ensuring the orderly streaming transmission of the response content.
It achieves coherent and consistent broadcasting of response content, avoids repetition and omissions, ensures that the client can identify the start and end boundaries of the broadcast, and improves the fluency and completeness of the broadcast.
Smart Images

Figure CN123075129A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of intelligent outbound real-time dialogue, and in particular to an intelligent outbound response broadcast control method and device. Background Technology
[0002] Intelligent outbound calling typically takes time to generate a complete outbound call script from a large model. If text-to-speech (TTS) is performed after all content is generated, the customer will hear a few seconds of silence. Directly handing over each model segment to TTS results in interruptions due to numbers, variables, or labels. Furthermore, the task graph needs to simultaneously transmit nodes, intents, slots, and ending states downstream; a simple text stream cannot meet these requirements.
[0003] When directly accessing the model stream, the following are some common issues.
[0004] 1. The first packet simultaneously handles initialization and broadcasting: downstream packets may play the structured response (answer) once, and then play it again from subsequent text streams, resulting in duplication.
[0005] 2. A single round of responses may come from multiple sources: knowledge answers, interface audio, entity clarification, and node guidance. Without a unified queue, it is easy to miss the latter half or reverse the order.
[0006] 3. Lack of clear ending after model flow interruption: Downstream keeps waiting for the next packet, and the final text sent to the database may not be consistent with what the customer actually heard. Summary of the Invention
[0007] The purpose of this application is to provide an intelligent outbound call response broadcast control method and device, which can realize the streaming and orderly broadcast of response content and improve the continuity and consistency of outbound call broadcast.
[0008] To achieve the above objectives, this application provides the following solution.
[0009] In a first aspect, this application provides an intelligent outbound call response broadcast control method, comprising: acquiring structured data of the current outbound call node; the structured data including node target, preset response text, external interface return data, customer intent, and slot instance; determining the response broadcast of the current outbound call node based on the structured data; processing the response broadcast according to preset rules to obtain segmented data packets; constructing a broadcast data packet sequence based on the segmented data packets, and sequentially sending all data packets in the broadcast data packet sequence to the client; the broadcast data packet sequence includes a start data packet, the segmented data packets, and an end data packet.
[0010] Secondly, this application provides an intelligent outbound call response broadcast control device, comprising: a data acquisition module for acquiring structured data of the current outbound call node; the structured data includes node target, preset response text, external interface return data, customer intent, and slot instance; a response generation module for determining the response broadcast of the current outbound call node based on the structured data; a data segmentation module for processing the response broadcast according to preset rules to obtain segmented data packets; and a broadcast sending module for constructing a broadcast data packet sequence based on the segmented data packets and sequentially sending all data packets in the broadcast data packet sequence to the client; the broadcast data packet sequence includes a start data packet, the segmented data packets, and an end data packet.
[0011] Based on the specific embodiments provided in this application, the following technical effects are disclosed.
[0012] This application provides an intelligent outbound call response broadcast control method and apparatus. By acquiring structured data of the current outbound call node, it can aggregate node configuration, business interface, and session identification information, providing a complete and unified input basis for response broadcast generation, ensuring that the response content is consistent with the current call node and customer session state. Based on the structured data, the response broadcast of the current outbound call node is determined, and broadcast content adapted to the current call interaction can be generated based on the current node information. The response broadcast is processed according to preset rules to obtain segmented data packets, which can split long texts into segments suitable for streaming transmission, completing the segmentation at appropriate semantic boundaries, improving the fluency of the broadcast. Based on the segmented data packets, a broadcast data packet sequence containing a start data packet, segmented data packets, and an end data packet is constructed and sequentially sent to the client, enabling ordered streaming push of the response content, allowing the client to identify the start and end boundaries of a round of broadcast, ensuring continuous and complete playback of the broadcast content. Attached Figure Description
[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0014] Figure 1 This is an application environment diagram of an intelligent outbound call response broadcast control method according to an embodiment of this application.
[0015] Figure 2 This is a flowchart illustrating an intelligent outbound call response broadcast control method provided in an embodiment of this application.
[0016] Figure 3This is a schematic diagram of the overall architecture of an intelligent outbound call response broadcast control method provided in an embodiment of this application.
[0017] Figure 4 This is a schematic diagram of the functional modules of an intelligent outbound call response broadcast control device provided in an embodiment of this application.
[0018] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation
[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0020] When intelligent outbound calls are directly connected to the model stream, the following problems still exist.
[0021] 1. Model segments are not playable segments: Phone numbers, amounts, English words, variable placeholders, and TTS tags may be split in half, causing pauses or misreading.
[0022] 2. Internal control tags may be leaked: If similar tags, residual prompts, or illegal audio commands generated by the model are directly issued, they may be heard by the customer or cause parsing failure.
[0023] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0024] The intelligent outbound call response broadcast control method provided in this application embodiment can be applied to, for example, Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up separately, integrated into server 104, or placed in the cloud or on another server. Terminal 102 can send the structured data of the current outbound call node to server 104. After receiving the structured data of the current outbound call node, server 104 determines the response broadcast and processes it to obtain segmented data packets. Based on the segmented data packets, combined with the start and end data packets, a broadcast data packet sequence is constructed, and finally, all data packets in the broadcast data packet sequence are sent to the client sequentially. Server 104 can provide feedback to terminal 102 with the obtained broadcast data packet sequence. Furthermore, in some embodiments, the intelligent outbound call response broadcast control method can also be implemented separately by server 104 or terminal 102. For example, terminal 102 can directly determine the response broadcast and process it to obtain segmented data packets based on the structured data of the current outbound call node. Based on the segmented data packets, combined with the start and end data packets, a broadcast data packet sequence is constructed, and finally, all data packets in the broadcast data packet sequence are sent to the client sequentially. Alternatively, server 104 can obtain the structured data of the current outbound node from the data storage system, determine the response broadcast and process it to obtain segmented data packets. Based on the segmented data packets, combined with the start data packet and the end data packet, a broadcast data packet sequence is constructed, and finally all data packets in the broadcast data packet sequence are sent to the client in sequence.
[0025] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.
[0026] In one exemplary embodiment, such as Figure 2 As shown, an intelligent outbound call response broadcast control method is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 204.
[0027] Step 201: Obtain the structured data of the current outbound call node. The structured data includes the node target, preset response text, external interface return data, customer intent, and slot instances. Specifically, the preset response text includes standard response scripts.
[0028] Step 202: Determine the response broadcast of the current outbound call node based on the structured data.
[0029] Step 203: Process the response broadcast according to preset rules to obtain segmented data packets.
[0030] Step 204: Based on the segmented data packets, construct a broadcast data packet sequence, and sequentially send all data packets in the broadcast data packet sequence to the client. The broadcast data packet sequence includes a start data packet, the segmented data packets, and an end data packet.
[0031] In another example, such as Figure 3 As shown, the intelligent outbound call response broadcast control method includes the following steps.
[0032] 1. Structured results and streaming source establishment.
[0033] First, obtain structured data such as node targets, preset response text, external interface return data, customer intent, slot instances, and response types according to the task diagram (i.e., outbound call dialogue process). These fields have been determined before the large model is generated below.
[0034] Among them, the node target, preset response text, response type and allowed event / slot range are mainly determined by the published task graph, the current node and node answer configuration; customer intent and slot instance are obtained from the basic identification and session state of this round; external interface returned data is obtained after the relevant interface node is executed.
[0035] 2. Response broadcast generation.
[0036] When intelligent responses or polishing are needed, a large model request is created based on recent conversations, node targets, hit knowledge, current / historical valid entities, external interface returned data, allowed business data, allowed data range, standard answers, and prompt words.
[0037] The recent dialogues, hit knowledge, current / historical valid entities, external interface returned data, and permitted business data are updated in real time with the current dialogue round; node targets, permitted data ranges, standard answers, and prompts are derived from the task graph and scenario configuration already loaded in the current session. This example creates a request snapshot of this data when creating a large model request.
[0038] 3. Construct the broadcast data packet sequence.
[0039] The broadcast data packet sequence includes the start packet (START), the segmented packet (DATA), and the end packet (END).
[0040] The START package carries data such as node target, customer intent, slot instance, response type, and large model status. It is used to open the playback channel and does not include the large model script that has not yet been completed.
[0041] The DATA packet only carries the newly added playable segments. Text already included in the previous packet will not be repeated in subsequent data packets.
[0042] The END packet indicates that the input for this round has been processed, and carries all the data sent in this round of response broadcast, the final large model state, and the task graph results.
[0043] The client uses START and END as status control packets and does not repeatedly broadcast the summary text in END.
[0044] In one example, the structured data also includes node configuration. The node configuration includes the response mode and model connection status of the current outbound call node. The response broadcast includes either a model response broadcast or a preset response text broadcast. Step 202 above can be replaced by steps 2021 and 2022.
[0045] Step 2021: Based on the node configuration, determine whether to call the large model to generate a response.
[0046] Step 2022: If the large model is invoked to generate a response, then the model response is broadcast based on the structured data and the large model. Otherwise, the preset response text is broadcast based on the preset response text.
[0047] In this example, the response mode is intelligent reply mode, and the online model service, intelligent reply, and related entity capabilities are enabled (i.e., the model is connected). Furthermore, if the current outbound call node is configured with a polishing flag, the intelligent reply can be further polished. When the response mode is not intelligent reply mode or the model call is abnormal (i.e., the model is disconnected), a preset response text is used.
[0048] When using a large model to generate responses, the large model returns an iterator or an asynchronous stream. The structured data R and the text stream I generated by the large model are combined into the same round of response; when no large model is needed, I is empty.
[0049] In this example, the protocol retains non-streaming fields, allowing older clients to still read preset response text even when streaming capabilities are disabled, thus avoiding the need to modify all call nodes during a single upgrade.
[0050] In one example, the response broadcast includes either a model response broadcast or a preset response text broadcast. The model response broadcast includes multiple sequentially generated model response segments. Step 203 above can be replaced by steps 2031 to 2033.
[0051] Step 2031: When the response broadcast is a model response broadcast, each model response segment is stored sequentially into the unsent buffer.
[0052] Step 2032: Using model broadcasting segmentation rules, the data in the unsent buffer is segmented to obtain the segmented data packets. The preset segmentation rules include at least one of a waiting time threshold, a data packet length threshold, a statement integrity index, and a number integrity index.
[0053] Step 2033: When the response broadcast is a preset response text broadcast, the preset response text broadcast is segmented using standard broadcast segmentation rules to obtain the segmented data packet. The standard broadcast segmentation rules include at least one of statement integrity indicators and number integrity indicators.
[0054] In this example, upon receiving a large model fragment, it is first appended to the unsent buffer, and the packet is directly segmented without counting the number of iterations. If the first fragment is too short or only contains a tag prefix, the system continues to wait for the next fragment.
[0055] Prioritize splitting at positions such as periods, question marks, exclamation marks, and commas; if there are no punctuation marks, splitting will only be performed when the character count or waiting time threshold is reached, and the character is not in the middle of numbers, English letters, variables, quotation marks, or tags.
[0056] In one example, there are multiple segmented data packets. Once any segmented data packet is generated, it is sent directly to the client.
[0057] In one example, prior to step 203 above, the intelligent outbound call response broadcast control method further includes: sequentially performing tag filtering and variable replacement on the response broadcast.
[0058] In this example, as Figure 3 As shown, internal prompt markers, Voice User Interface (VUI) prefixes, and unauthorized TTS tags in the large model text have been removed. The large model cannot directly create call control commands such as audio playback, call transfer, or hang-up.
[0059] Valid tags already configured in the preset response text are retained according to the whitelist. Interface variables, session parameters, and slots are replaced on the system side; missing values are deleted according to field configuration, default values are used, or the entire segment is switched to a fallback script.
[0060] In one example, the number of data packets to be segmented is multiple.
[0061] The broadcast data packet sequence corresponds to a unique round identifier, and each segmented data packet in the broadcast data packet sequence corresponds to a distinct sequence number. The client performs data deduplication based on the round identifier and the sequence number corresponding to each segmented data packet.
[0062] In this example, each round of outbound call response generates a unique round identifier and sequence number. The client removes duplicates based on the session identifier, round identifier, and sequence number, and data that is late in the old round is no longer played.
[0063] In one example, the response broadcast includes a baseline script. The baseline script is pre-parsed based on the outbound dialogue flow corresponding to the current outbound call node. The baseline script includes knowledge response scripts, node scripts, entity clarification scripts, and interface audio.
[0064] In this example, as Figure 3 As shown, a baseline script (multi-source script queue) is established by parsing the task graph (i.e., the outbound call dialogue process). The baseline script includes knowledge response, node script, entity clarification, interface audio, and model script. Fixed risk warnings in the response broadcast are not overwritten by the output of the large model; the next element data is sent only after the previous element is sent, and all element data share the same round sequence number.
[0065] In one example, the broadcast data packet sequence also includes error data packets. The error data packets include an error code and a content playback status. The error code is used to identify the error type. The content playback status is used to indicate whether the segmented data packets have been sent to the client.
[0066] In this example, as Figure 3 As shown, error packets (ERROR) are used in situations where streaming, protocol encapsulation, or content processing cannot be recovered, including error codes and whether the content has already been played.
[0067] In one example, the number of segmented data packets is multiple. When the data packets in the broadcast data packet sequence are sent sequentially to the client, each segmented data packet sent to the client is collected in real time, and based on all the segmented data packets sent to the client, the total data sent in this round of response broadcasting is determined.
[0068] In this example, as Figure 3 As shown, for each DATA packet sent, the actual data packet sent is appended to the cumulative value A. If the model stream has not ended, the next segment is read, and the process returns to the "segment buffering and cleanup, boundary judgment" step. The fragment is then merged with the remaining content in the unsent buffer and a second judgment is made to determine whether a playable segment has been formed. Both the data storage and the END packet are based on A, and the uncleaned original model text is no longer used.
[0069] When a large model terminates normally, the remaining text in the buffer is processed first, and then the END message is sent. When there is no stream but structured data exists, the non-streaming preset response text is used to terminate the stream directly.
[0070] If an error occurs before the first DATA packet, a complete preset response text can be sent as a fallback. After some content has been broadcast, the preset response text will not be played from the beginning; instead, a short closing sequence will be added or the content will end abruptly.
[0071] When the maximum total duration or segment idle time is reached, reading is actively stopped, a timeout status is written, and an ERROR or END is sent to avoid downstream users waiting indefinitely.
[0072] If the large model service is not deployed, the capability switch is off, or the request fails, the text stream is skipped, and the original task graph's precise dialogue, rejection dialogue, and node jumps are still executed according to the old logic.
[0073] This application also provides an application scenario in which the above-described intelligent outbound call response broadcast control method is applied. Specifically, the intelligent outbound call response broadcast control method provided in this embodiment can be applied in an intelligent outbound call interaction scenario. The intelligent outbound call interaction scenario includes an outbound call node scheduling stage, an response broadcast processing link, and a client broadcast playback stage. The outbound call task enters the response broadcast processing link from the outbound call node scheduling stage, obtains the corresponding broadcast data through structured data parsing and data packet processing, and then enters the downstream client broadcast playback stage. The intelligent outbound call response broadcast control method provided in this embodiment belongs to the broadcast data generation and distribution stage in the response broadcast processing link. Specifically, in the intelligent outbound call response broadcast processing link, response broadcast determination, data packet segmentation, and sequence construction can be completed based on node structured data, achieving standardized and orderly outbound call voice response broadcast distribution.
[0074] Based on the same inventive concept, this application also provides an intelligent outbound call response broadcast control device for implementing the intelligent outbound call response broadcast control method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations of one or more intelligent outbound call response broadcast control device embodiments provided below can be found in the limitations of the intelligent outbound call response broadcast control method described above, and will not be repeated here.
[0075] In one exemplary embodiment, such as Figure 4 As shown, an intelligent outbound call response broadcast control device is provided, including a data acquisition module 001, a response generation module 002, a data segmentation module 003, and a broadcast sending module 004.
[0076] The data acquisition module 001 is used to acquire the structured data of the current outbound call node. The structured data includes node target, preset response text, external interface return data, customer intent, and slot instances.
[0077] The response generation module 002 is used to determine the response broadcast of the current outbound call node based on the structured data.
[0078] The data segmentation module 003 is used to process the response broadcast according to preset rules to obtain segmented data packets.
[0079] The broadcast sending module 004 is used to construct a broadcast data packet sequence based on the segmented data packets, and sequentially send all data packets in the broadcast data packet sequence to the client. The broadcast data packet sequence includes a start data packet, the segmented data packets, and an end data packet.
[0080] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 5 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores structured data for the current outbound call node. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements an intelligent outbound call response broadcast control method.
[0081] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer equipment to which the present application is applied. Specific computer equipment may include, for example, [the following is a list of possible additional structures]. Figure 5 The diagram shows more or fewer components, or combinations of certain components, or different component arrangements.
[0082] This application addresses the problem that large models output text in segments, while outbound calling platforms need to continuously broadcast text with complete semantics and maintain node states.
[0083] This application does not wait for the large model to return the entire script at once. Instead, it combines the structured data and iterative text stream in the outbound call dialogue flow task graph and transmits it downstream through four types of data packets: START, DATA, END, and ERROR, separating the initialization state, incremental text, summary results, and abnormal states. Before sending, it also processes segment boundaries, internal tags, variables, and the playback order of multiple segments. Using the unsent buffer and the cumulative sent value, it selects playable segmentation points based on punctuation, length, waiting time, numerical continuity, variable and tag boundaries. This application distinguishes between the large model text and the legal control instructions in the task graph, cleansing internal tags, performing variable replacement and TTS instruction supplementation, and providing deterministic fallback for missing variables. This application establishes an ordered playback queue for scripts from multiple sources, filling the final script with actually sent segments, and handling empty streams, timeouts, and mid-stream anomalies based on the played content.
[0084] The applicable scenarios for this application are as follows.
[0085] 1. Financial, insurance, and securities follow-up: Dynamically answer customer inquiries while ensuring that risk warnings, compliance scripts, and subsequent guidance are played in the original order.
[0086] 2. Government, medical and education reminders: When the model generates a longer explanation, it will first broadcast the first half of the sentence that is already complete, reducing the silent waiting during the phone call.
[0087] 3. E-commerce and customer care: Securely replace interface variables such as customer name, order, and time into streaming segments, and switch to preset scripts when variables are missing.
[0088] 4. Online voice customer service: Knowledge answers, node scripts, interface audio, and model content are sent continuously in a queue to avoid playing only the first segment.
[0089] 5. Compatibility with modified scenarios: When the model is not enabled or is unavailable, the same start and end protocols are still used, and the downstream continues to play the original precise script.
[0090] This application uses the structured dialogue result as the fixed state of the current outbound call node and the large model iterator as the incremental text source. The start data packet is only responsible for initialization, the segmented data packet only sends newly added and playable content, and the end data packet refills the complete script that has actually been sent. In case of an error, a preset response text, a closing script, or an error packet is selected based on whether content has already been broadcast. Specifically, the beneficial effects of this application are as follows.
[0091] 1. Customers hear the first complete segment earlier.
[0092] This application only waits for the formation of the first safe broadcast boundary, without waiting for the large model to generate the entire speech. The waiting time for the first segment is shortened from the full generation time to the generation time of the first few segments.
[0093] If the first segment is too short, it should be buffered first to avoid directly handing over half a word or tag prefix to TTS in order to pursue the speed of the first packet.
[0094] 2. Streaming playback will not repeat the first segment or miss the later segment.
[0095] START only initializes the system, DATA only stores newly added text, and END only summarizes the status, clearly defining the downstream playback responsibilities.
[0096] Scripts from multiple sources are uniformly entered into an ordered queue, and knowledge answers, interface audio, and node guidance are sent sequentially according to their serial numbers.
[0097] 3. Keep numbers, variables, and voice labels intact.
[0098] Before splitting, check the continuity of numbers, variable placeholders, and label boundaries. Phone numbers, amounts, and dates will not be split due to model fragmentation.
[0099] Illegal tags are removed from this application, and legitimate TTS or audio commands are generated from a whitelist, so the customer will not hear internal control characters.
[0100] 4. The actual broadcast content is consistent with the database result.
[0101] The cumulative value only appends the text that was actually sent successfully. This value is used for both END and session round records. The debug page can restore the order in which the customer actually heard the text.
[0102] The original model output can still be saved separately as controlled debugging information, but it will not overwrite the broadcast results after purification and variable replacement.
[0103] 5. The current round can be clearly ended when the flow is interrupted.
[0104] Empty flow, timeout, and mid-process exception all have clear ERROR or END, and the calling platform does not need to rely on connection disconnection to guess whether the current round has ended.
[0105] In case of an anomaly, the system selects either a full preset response text or a brief closing statement based on the already played content, avoiding both prolonged silence and repeating the already played content from the beginning.
[0106] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0107] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).
[0108] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0109] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0110] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for controlling intelligent outbound call response broadcasting, characterized in that, The method includes: Obtain the structured data of the current outbound call node; the structured data includes node target, preset response text, external interface return data, customer intent, and slot instance; Based on the structured data, determine the response broadcast of the current outbound call node; The response broadcast is processed according to preset rules to obtain segmented data packets; Based on the segmented data packets, a broadcast data packet sequence is constructed, and all data packets in the broadcast data packet sequence are sent to the client in sequence; the broadcast data packet sequence includes a start data packet, the segmented data packets, and an end data packet.
2. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, The structured data also includes node configuration; the node configuration includes the response mode and model connection status of the current outbound call node; the response broadcast includes either model response broadcast or preset response text broadcast; Based on the structured data, the response broadcast of the current outbound call node is determined, specifically including: Based on the node configuration, determine whether to invoke the large model to generate a response; If a large model is invoked to generate a response, the model response is broadcast based on the structured data and the large model; otherwise, the preset response text is broadcast based on the preset response text.
3. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, The response broadcast includes either a model response broadcast or a preset response text broadcast; the model response broadcast includes multiple sequentially generated model response segments. The response broadcast is processed according to preset rules to obtain segmented data packets, specifically including: When the response broadcast is a model response broadcast, each model response segment is stored sequentially into the unsent buffer; The data in the unsent buffer is segmented using model broadcasting segmentation rules to obtain the segmented data packets; the model broadcasting segmentation rules include at least one of waiting time threshold, data packet length threshold, statement integrity index, and number integrity index. When the response broadcast is a preset response text broadcast, the preset response text broadcast is segmented using standard broadcast segmentation rules to obtain the segmented data packet; the standard broadcast segmentation rules include at least one of statement integrity indicators and number integrity indicators.
4. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, The number of segmented data packets is multiple; once any segmented data packet is generated, it is sent directly to the client.
5. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, Before processing the response broadcast according to preset rules to obtain segmented data packets, the intelligent outbound call response broadcast control method further includes: sequentially performing tag filtering and variable replacement on the response broadcast.
6. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, The number of data packets to be segmented is multiple; The broadcast data packet sequence corresponds to a unique round identifier, and each segmented data packet in the broadcast data packet sequence corresponds to a different sequence number; the client performs data deduplication based on the round identifier and the sequence number corresponding to each segmented data packet.
7. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, The response broadcast includes a standard script; The baseline script is pre-analyzed based on the outbound dialogue flow corresponding to the current outbound call node; The benchmark scripts include knowledge response scripts, node scripts, entity clarification scripts, and interface audio.
8. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, The broadcast data packet sequence also includes error data packets; the error data packets include error codes and content playback status. The error code is used to mark the error type; the content playback status is used to mark whether the segmented data packet has been sent to the client.
9. The intelligent outbound call response broadcast control method according to claim 1, characterized in that, The number of data packets to be segmented is multiple; When the data packets in the broadcast data packet sequence are sent to the client in sequence, each segmented data packet sent to the client is collected in real time, and all data to be sent in this round of response broadcast is determined based on all the segmented data packets sent to the client.
10. An intelligent outbound call response broadcast control device, characterized in that, The device includes: The data acquisition module is used to acquire the structured data of the current outbound call node; the structured data includes node target, preset response text, external interface return data, customer intent and slot instance; The response generation module is used to determine the response broadcast of the current outbound call node based on the structured data; The data segmentation module is used to process the response broadcast according to preset rules to obtain segmented data packets; The broadcast sending module is used to construct a broadcast data packet sequence based on the segmented data packets, and sequentially send all data packets in the broadcast data packet sequence to the client; the broadcast data packet sequence includes a start data packet, the segmented data packets, and an end data packet.
