Map voice broadcasting method, device and equipment based on large model and storage medium

Abstract

The invention relates to the technical field of voice broadcasting, and discloses a map voice broadcasting method and device based on a large model, equipment and a storage medium. The method comprises the following steps: a server generates original broadcast copywriting of a plurality of broadcast positions in current navigation and inputs the original broadcast copywriting into a large model; the large model generates a large model broadcast copywriting of each broadcast position and returns the large model broadcast copywriting to the server; the server side obtains a score value of historical navigation, generates a switching identifier, and sends the large model broadcast copywriting, the original broadcast copywriting and the switching identifier to the client side; the client determines an actual broadcast copywriting and actual broadcast information of each broadcast position, calculates an actual broadcast score and a large model broadcast score of each broadcast position, and sends the actual broadcast score and the large model broadcast score of each broadcast position to the server; and the server side calculates the overall broadcast score of the navigation. According to the application, the accuracy, the timeliness and the individuation level of broadcasting are improved, the flexibility and the adaptability of the device are enhanced, and different user requirements are met.

Description

Technical Field

[0001] The present invention relates to the technical field of voice broadcast, and particularly to a map voice broadcast method, device, equipment and storage medium based on a large model. Background Art

[0002] Currently, navigation broadcast copywriting is generally generated by the server or in the form of a configuration file. Multiple broadcast copywritings are generated at the same location, and different speed levels correspond to different copywritings. Each copywriting has its own broadcast interval, and then it is sent to the client. When the client actually runs, different copywritings are matched according to the current speed.

[0003] However, when the server generates navigation broadcast copywriting, the client requests the server copywriting before each navigation. The server needs to update the code to update the broadcast copywriting and timing, and the process is cumbersome and time-consuming. When generating navigation broadcast copywriting in the form of a configuration file, although the update efficiency is improved and only the file needs to be modified without recompiling the server, it is based on fixed rules and cannot adapt to different driving habits and real-time road conditions. The generated copywriting lacks personalization and real-time adaptability. These two methods are difficult to meet the high requirements of navigation broadcast for timeliness, personalization and accuracy. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to overcome the deficiencies in the prior art and provide a map voice broadcast method, device, equipment and storage medium based on a large model.

[0005] The present invention provides the following technical solutions: In a first aspect, an embodiment of the present disclosure provides a map voice broadcast method based on a large model, and the method includes: The server generates original broadcast copywriting for multiple broadcast positions in this navigation, and inputs each of the original broadcast copywritings into the large model; The large model generates large model broadcast copywriting for each of the broadcast positions, and returns each of the large model broadcast copywritings to the server; The server obtains the scoring value of the historical navigation, generates a switching identifier according to the scoring value of the historical navigation, and sends each of the large model broadcast copywritings, each of the original broadcast copywritings and the switching identifier to the client; The client determines the actual broadcast copywriting for each of the broadcast positions in this navigation according to the switching identifier, records the corresponding actual broadcast information, calculates the actual broadcast score for each of the broadcast positions in this navigation according to each of the actual broadcast information, and simulates and calculates the large model broadcast score for each of the broadcast positions in this navigation, and sends the actual broadcast score and the large model broadcast score for each of the broadcast positions in this navigation to the server; The server calculates the overall broadcast score of the current navigation based on the actual broadcast scores and the large model broadcast scores of each of the broadcast positions in the current navigation.

[0006] In an alternative embodiment, the switching identifier includes a first switching identifier and a second switching identifier. Obtaining the scoring value of the historical navigation and generating a switching identifier based on the scoring value of the historical navigation includes: Obtain the scoring values of the previous two historical navigations, calculate the change value of the scoring values of the previous two historical navigations, and determine whether the change value is less than a preset convergence threshold; If the change value is less than the preset convergence threshold, use each of the large model broadcast texts as the actual broadcast text for the corresponding broadcast position in the current navigation, and generate the first switching identifier; If the change value is greater than or equal to the preset convergence threshold, use each of the original broadcast texts as the actual broadcast text for the corresponding broadcast position in the current navigation, and generate the second switching identifier.

[0007] In an alternative embodiment, determining the actual broadcast text for each of the broadcast positions in the current navigation according to the switching identifier and recording the corresponding actual broadcast information includes: Determine the actual broadcast text for each of the broadcast positions in the current navigation according to the first switching identifier or the second switching identifier; Record the actual broadcast information when each of the broadcast positions in the current navigation broadcasts the corresponding actual broadcast text, including the actual broadcast position, the best broadcast position, the broadcast upper limit position, the broadcast lower limit position, and the actual broadcast completion position.

[0008] In an alternative embodiment, calculating the actual broadcast score for each of the broadcast positions in the current navigation according to each of the actual broadcast information includes: Using a preset broadcast score calculation formula, calculate the actual broadcast score for each of the broadcast positions in the current navigation according to the actual broadcast position, the best broadcast position, the broadcast upper limit position, the broadcast lower limit position, and the actual broadcast completion position in each of the actual broadcast information; Wherein, the preset broadcast score calculation formula is:

[0009] In the formula, is the actual broadcast score of the i-th broadcast position, is the actual broadcast position of the i-th broadcast position, is the best broadcast position of the i-th broadcast position, is the broadcast upper limit position of the i-th broadcast position, is the broadcast lower limit position of the i-th broadcast position, is the actual broadcast completion position of the i-th broadcast position.

[0010] In an alternative embodiment, calculating the overall broadcast score of the current navigation based on the actual broadcast scores and the large model broadcast scores of each of the broadcast positions in the current navigation includes: Calculating the sum of the large model broadcast scores of all the broadcast positions, calculating the sum of the actual broadcast scores of all the broadcast positions, and obtaining the total number of the broadcast positions; Using a preset overall broadcast score calculation formula, calculating the overall broadcast score of the current navigation based on the sum of the large model broadcast scores of all the broadcast positions, the sum of the actual broadcast scores of all the broadcast positions, and the total number of the broadcast positions; wherein, the preset overall broadcast score calculation formula is:

[0011] In the formula, is the overall broadcast score of the current navigation, is the total number of the broadcast positions, is the large model broadcast score of the i-th broadcast position, is the sum of the large model broadcast scores of all the broadcast positions, is the actual broadcast score of the i-th broadcast position, is the sum of the actual broadcast scores of all the broadcast positions.

[0012] In an alternative embodiment, the method further includes: The client transmits the actual broadcast copywriting, the actual broadcast score, and the large model broadcast score of each of the broadcast positions in the current navigation to the server; The server inputs the actual broadcast copywriting, the actual broadcast score, and the large model broadcast score of each of the broadcast positions in the current navigation into the large model; The large model corrects the corresponding actual broadcast copywriting according to the actual broadcast scores and the large model broadcast scores of each of the broadcast positions in the current navigation to obtain a plurality of optimized broadcast copywritings, and returns each of the optimized broadcast copywritings to the server; The server simulates and calculates the optimized broadcast scores of each of the optimized broadcast copywritings, and feeds back the optimal broadcast copywriting with the highest optimized broadcast score to the large model; The large model is trained according to the optimal broadcast copywriting to obtain an optimized large model.

[0013] In an alternative embodiment, correcting the corresponding actual broadcast copywriting according to the actual broadcast scores and the large model broadcast scores of each of the broadcast positions in the current navigation to obtain a plurality of optimized broadcast copywritings includes: Determine whether the actual broadcast scores of each of the described broadcast positions in this navigation are greater than a preset score threshold, and determine whether the large model broadcast scores of each of the described broadcast positions in this navigation are greater than the preset score threshold; If the actual broadcast scores of each of the described broadcast positions in this navigation are greater than the preset score threshold, and / or the large model broadcast scores of each of the described broadcast positions in this navigation are greater than the preset score threshold, then use a preset correction rule to correct each of the actual broadcast copywritings to obtain multiple optimized broadcast copywritings.

[0014] In a second aspect, an embodiment of the present disclosure provides a map voice broadcast device based on a large model, and the device includes a server, a large model, and a client; The server is configured to generate original broadcast copywritings for multiple broadcast positions in this navigation, and input each of the original broadcast copywritings into the large model; The large model is configured to generate large model broadcast copywritings for each of the broadcast positions, and return each of the large model broadcast copywritings to the server; The server is further configured to obtain a scoring value of a historical navigation, generate a switching identifier according to the scoring value of the historical navigation, and send each of the large model broadcast copywritings, each of the original broadcast copywritings, and the switching identifier to the client; The client is configured to determine the actual broadcast copywritings for each of the broadcast positions in this navigation according to the switching identifier, record the corresponding actual broadcast information, calculate the actual broadcast scores for each of the broadcast positions in this navigation according to each of the actual broadcast information, and simulate and calculate the large model broadcast scores for each of the broadcast positions in this navigation, and send the actual broadcast scores and large model broadcast scores for each of the broadcast positions in this navigation to the server; The server is further configured to calculate an overall broadcast score of this navigation according to the actual broadcast scores and large model broadcast scores for each of the broadcast positions in this navigation.

[0015] In a third aspect, an embodiment of the present disclosure provides a computer device, and the computer device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps of the method for map voice broadcast based on a large model described in the first aspect are implemented.

[0016] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for map voice broadcast based on a large model described in the first aspect are implemented.

[0017] Advantages of this application: The map voice broadcast method based on a large model provided by the embodiments of the present application trains the large model by collecting information such as the difference between the actual broadcast position and the expected broadcast position, the road level, speed, and turning model of the broadcast position multiple times, generates a broadcast copy that conforms to the road conditions at a certain place, provides a corresponding discrimination algorithm to switch between the generated broadcast copy and the original broadcast copy, and also provides relevant calculation formulas to verify the effect of applying the broadcast copy of the large model to actual broadcasts. This not only improves the accuracy, timeliness, and personalization level of broadcasts, enhances the flexibility and adaptability of the device, reduces the development cost, but also can dynamically adjust the broadcast strategy according to the actual road conditions and driving habits to ensure that the broadcast content is accurate and the broadcast timing is appropriate, meeting the needs of different users.

[0018] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, makes the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other relevant drawings can also be obtained based on these drawings. In each drawing, similar components are numbered similarly.

[0020] Figure 1 Shows a flowchart of a map voice broadcast method based on a large model provided by the embodiments of the present application; Figure 2 Shows a schematic structural diagram of a map voice broadcast device provided by the embodiments of the present application; Figure 3 Shows a schematic structural diagram of a computer device provided by the embodiments of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The following details the embodiments of the present invention. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions from beginning to end. The embodiments described below with reference to the drawings are exemplary and are only used to explain the present invention and should not be construed as limiting the present invention.

[0022] It should be noted that the terms "first" and "second" are only used for descriptive purposes and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality of" means two or more, unless otherwise specifically defined.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs. The terms used in the specification of the template herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0024] Embodiment 1 As Figure 1 shown, it is a flowchart of a map voice broadcast method based on a large model in an embodiment of the present application. The map voice broadcast method based on a large model provided by the embodiment of the present application includes the following steps: Step S110, the server generates the original broadcast copywriting for multiple broadcast positions in this navigation and inputs each original broadcast copywriting into the large model.

[0025] In this embodiment, first, the server generates the original broadcast copywriting for multiple broadcast positions in this navigation according to normal logic. The original broadcast copywriting includes broadcast statements, the best broadcast positions, the road grades of the navigation segments, the preset speeds corresponding to these road grades (empirical values, generally 80 km / h for urban expressways and 120 km / h for highways, etc.), the actual speeds of the roads at these positions (empirical values of all vehicles passing through this section of the road in history, collected by the server), steering models and other information.

[0026] After the server generates the original broadcast copywriting for multiple broadcast positions, it inputs each original broadcast copywriting into the large model.

[0027] The above method generates the original broadcast copywriting by the server and inputs it into the large model, providing basic data and corpus support for the subsequent generation of the broadcast copywriting by the large model. Various information in the original broadcast copywriting, such as broadcast statements, the best broadcast positions, road grades and corresponding speeds, etc., can help the large model better understand the broadcast scenarios and requirements, thus laying a foundation for generating broadcast copywriting that is more in line with the actual road conditions and driving habits.

[0028] Step S120, the large model generates the large model broadcast copywriting for each broadcast position and returns each large model broadcast copywriting to the server.

[0029] Preferably, the structure of the large model includes a base model + a knowledge base. After receiving multiple original broadcast copywriting sent by the server, each original broadcast copywriting is used as the corpus input of the knowledge base and added to the knowledge base through a basic embedding model. When using the large model, it can be retrieved through the RAG (Retrieval-Augmented Generation) technology, and the accuracy of the content generated by the large model is enhanced by retrieving the external knowledge base. Its core process includes: Retrieval stage: Screen relevant data from the knowledge base; Generation stage: Optimize the model output in combination with the retrieval results to reduce the "hallucination" problem.

[0030] Through the above process, the large model generates corresponding large model broadcast copywriting for each broadcast position and returns each large model broadcast copywriting to the server.

[0031] In the above method, the large model generates large model broadcast copywriting based on the original broadcast copywriting, which can realize the optimization and personalized generation of the broadcast copywriting. By using the structure of the base model + knowledge base and the RAG technology, the accuracy and relevance of the generated content can be enhanced, the hallucination problem can be reduced, the quality and practicality of the broadcast copywriting can be improved, and it can better meet the actual road conditions and the personalized needs of drivers.

[0032] Step S130, the server obtains the scoring value of the historical navigation, generates a switching identifier according to the scoring value of the historical navigation, and sends each large model broadcast copywriting, each original broadcast copywriting, and the switching identifier to the client.

[0033] It can be understood that the server can obtain the scoring values of the previous two historical navigations, calculate the change value of the scoring values of the previous two historical navigations, and determine whether the change value is less than a preset convergence threshold. If the change value is less than the preset convergence threshold, each large model broadcast copywriting is used as the actual broadcast copywriting for the corresponding broadcast position in this navigation, and a first switching identifier is generated; if the change value is greater than or equal to the preset convergence threshold, each original broadcast copywriting is used as the actual broadcast copywriting for the corresponding broadcast position in this navigation, and a second switching identifier is generated.

[0034] It should be noted that if this voice broadcast method of the present application is used for the first time or the second time, the scoring values of the previous two historical navigations can be defaulted to 0, and the change value of the scoring values of the previous two historical navigations is also 0.

[0035] After generating the corresponding switching identifier, each large model broadcast copywriting, each original broadcast copywriting, and the first switching identifier or the second switching identifier are sent to the client.

[0036] In the above method, the server generates a switching identifier based on the scoring value of historical navigation and sends relevant data to the client, realizing the control of dynamic selection of the broadcast copywriting. This judgment mechanism based on historical data can reasonably switch between the large model broadcast copywriting and the original broadcast copywriting, ensuring the accuracy and reliability of the current navigation broadcast copywriting, and at the same time providing a clear execution basis for the client.

[0037] Step S140, the client determines the actual broadcast copywriting for each broadcast position in the current navigation according to the switching identifier, records the corresponding actual broadcast information, calculates the actual broadcast score for each broadcast position in the current navigation according to each actual broadcast information, and simulates and calculates the large model broadcast score for each broadcast position in the current navigation, and sends the actual broadcast score and the large model broadcast score for each broadcast position in the current navigation to the server.

[0038] Specifically, after receiving the first switching identifier or the second switching identifier, the client determines the actual broadcast copywriting for each broadcast position in the current navigation according to the first switching identifier or the second switching identifier. If it is the first switching identifier, each large model broadcast copywriting is used as the actual broadcast copywriting for the corresponding broadcast position in the current navigation; if it is the second switching identifier, each original broadcast copywriting is used as the actual broadcast copywriting for the corresponding broadcast position in the current navigation.

[0039] When each broadcast position broadcasts according to the actual broadcast copywriting, record the current actual broadcast information, including the actual broadcast statement, the actual broadcast position, the best broadcast position, the broadcast upper limit position, the broadcast lower limit position, the actual broadcast completion position, the road grade of the navigation section, the steering model, the difference between the actual speed and the preset speed of the section, etc.

[0040] It should be noted that if the navigation voice cannot be broadcast at the expected position, the biggest influencing factor is the conflict of the broadcast statement. If there is no conflict, this statement can basically be broadcast at the expected position. Therefore, not only the starting position of the broadcast needs to be considered, but also the position when the broadcast is completed needs to be considered, because this position will affect the next broadcast statement.

[0041] Furthermore, using the preset broadcast score calculation formula, according to the actual broadcast position, the best broadcast position, the broadcast upper limit position, the broadcast lower limit position, and the actual broadcast completion position in each actual broadcast information, calculate the actual broadcast score for each broadcast position in the current navigation. Among them, the preset broadcast score calculation formula is:

[0042] In the formula, is the actual broadcast score of the i-th broadcast position, is the actual broadcast position of the i-th broadcast position, is the optimal broadcast position for the i-th broadcast position, is the upper limit position of the broadcast for the i-th broadcast position, is the lower limit position of the broadcast for the i-th broadcast position, is the actual broadcast completion position for the i-th broadcast position.

[0043] Understandably, in the above preset broadcast score calculation formula, the actual broadcast score The lower it is, the better the broadcast effect. The meaning of the formula is that the closer the actual broadcast position is to the optimal broadcast position, the more likely it is to fall within the lower limit position of the broadcast after completion inside, the better the broadcast effect; the smaller the upper and lower interval between the upper limit position and the lower limit position of the broadcast, the greater the cost of movement, and for the same moving distance is greater, and the effect is not good.

[0044] It should be noted that if the actual broadcast completion position lags behind the lower limit position of the broadcast, that is , it means that the broadcast is completed too late, which may have exceeded the effective guidance range and reduced the reference value for the driver, belonging to a broadcast error. To accurately evaluate the broadcast effect, when the above lag occurs, the penalty mechanism will directly adjust what may originally be a negative value or a small positive value directly to . This adjustment increases the error impact of this position in the calculation, making the score more intuitively reflect the broadcast lag problem. This penalty mechanism strengthens the requirement for the accuracy of the broadcast completion position, prompting the device to pay more attention to improving the timeliness and accuracy of the broadcast completion during optimization.

[0045] Similarly, using the above preset broadcast score calculation formula and based on the large model broadcast copy generated previously, simulate and calculate the large model broadcast scores of each broadcast position in this navigation , and send the actual broadcast scores and large model broadcast scores of each broadcast position in this navigation to the server.

[0046] In the above method, the client determines the actual broadcast copy according to the switching identifier, records the broadcast information, calculates the score, and can quantitatively evaluate the broadcast effect of this navigation. By recording detailed actual broadcast information, various situations during the broadcast process can be comprehensively understood, and calculating the actual broadcast score and the large model broadcast score provides important data support for subsequent model optimization and server-side strategy adjustment, contributing to the formation of a continuous improvement loop.

[0047] Step S150, the server calculates the overall broadcast score of this navigation according to the actual broadcast scores and large model broadcast scores of each broadcast position in this navigation.

[0048] Finally, the server calculates the sum of the large model broadcast scores at all broadcast positions, calculates the sum of the actual broadcast scores at all broadcast positions, and obtains the total number of broadcast positions. Using a preset overall broadcast score calculation formula, based on the sum of the large model broadcast scores at all broadcast positions, the sum of the actual broadcast scores, and the total number of broadcast positions, the overall broadcast score for this navigation is calculated. The preset overall broadcast score calculation formula is as follows:

[0049] In the formula, is the overall broadcast score for this navigation, is the total number of broadcast positions, is the large model broadcast score at the i-th broadcast position, is the sum of the large model broadcast scores at all broadcast positions, is the actual broadcast score at the i-th broadcast position, is the sum of the actual broadcast scores at all broadcast positions.

[0050] In the above method, the server calculates the overall broadcast score for this navigation, which can comprehensively evaluate the overall effect of this navigation broadcast and provide a key basis for subsequent adjustment of the broadcast strategy and optimization of the model. Through the overall broadcast score, the performance of the broadcast device can be intuitively understood, and then the server can be guided to conduct more targeted training and optimization of the large model.

[0051] In an alternative embodiment, the map voice broadcast method based on a large model in this embodiment further includes: the client transmits the actual broadcast copywriting, actual broadcast scores, and large model broadcast scores at each broadcast position in this navigation back to the server, and the server then inputs the actual broadcast copywriting, actual broadcast scores, and large model broadcast scores at each broadcast position in this navigation into the large model; the large model corrects the corresponding actual broadcast copywriting according to the actual broadcast scores and large model broadcast scores at each broadcast position in this navigation to obtain multiple optimized broadcast copywritings, and returns each optimized broadcast copywriting to the server; the server uses the previous preset broadcast score calculation formula to simulate and calculate the optimized broadcast scores of each optimized broadcast copywriting, and feeds back the optimal broadcast copywriting with the highest optimized broadcast score to the large model. Based on this optimal broadcast copywriting, the large model can carry out targeted training while keeping its original structure and parameters stable, achieving self-optimization and improvement, so that the optimized large model can generate a more accurate and demand-compliant voice broadcast plan according to the characteristics and advantages of the learned optimal broadcast copywriting in subsequent voice broadcast work, thereby providing a better and more accurate voice broadcast service for users.

[0052] Understandably, the method of correcting the corresponding actual broadcast copy based on the actual broadcast scores and the large model broadcast scores at each broadcast position in the current navigation to obtain multiple optimized broadcast copies may include: determining whether the actual broadcast scores at each broadcast position in the current navigation are greater than a preset score threshold, and determining whether the large model broadcast scores at each broadcast position in the current navigation are greater than the preset score threshold; if the actual broadcast scores at each broadcast position in the current navigation are greater than the preset score threshold, and / or the large model broadcast scores at each broadcast position in the current navigation are greater than the preset score threshold, it proves that the actual broadcast scores and the large model broadcast scores at each broadcast position in the current navigation are relatively high, and it is necessary to correct each actual broadcast copy using a preset correction rule to obtain multiple optimized broadcast copies. Among them, the preset correction rule includes but is not limited to the following rules: (1) If the difference between the actual speed and the preset speed of the road section is positive and the difference is greater than a certain value (which can be determined according to the actual situation), the length of the actual broadcast statement is shortened; if the difference between the actual speed and the preset speed of the road section is negative and the difference is less than a certain value (which can be determined according to the actual situation), the length of the actual broadcast statement is increased; (2) Generate appropriate actual broadcast copies for various steering models; (3) If the actual broadcast position is a relatively long distance before the optimal broadcast position (the distance can be determined according to the actual situation), it proves that the broadcast is advanced. At this time, the length of the actual broadcast statement is increased to make the prompt clearer; if the actual broadcast position is a relatively long distance after the optimal broadcast position (the distance can be determined according to the actual situation), it proves that the broadcast is delayed. At this time, the length of the actual broadcast statement is shortened to make the prompt more timely.

[0053] It should be noted that the preset correction rule can be adjusted according to the actual situation, and this embodiment does not limit it.

[0054] In the above method, based on the actual broadcast data transmitted back by the client, the large model can automatically correct low-score copies (such as adjusting the statement length and steering prompts), forming a closed loop of "generation - evaluation - optimization". By continuously learning the actual broadcast effect, the large model gradually optimizes the generation strategy, realizes self-improvement, continuously improves the broadcast quality, better meets the user's needs, and provides better services for subsequent navigation broadcasts.

[0055] The map voice broadcast method based on a large model provided by the embodiments of the present application trains the large model by collecting information such as the difference between the actual broadcast position and the expected broadcast position, the road level, speed, and turning model of the broadcast position multiple times, generates a broadcast copy that conforms to the road conditions at a certain place, provides a corresponding discrimination algorithm to switch between the generated broadcast copy and the original broadcast copy, and also provides relevant calculation formulas to verify the effect of applying the broadcast copy of the large model to actual broadcasts. This not only improves the accuracy, timeliness, and personalization level of broadcasts, enhances the flexibility and adaptability of the device, reduces the development cost, but also can dynamically adjust the broadcast strategy according to the actual road conditions and driving habits to ensure that the broadcast content is accurate and the broadcast timing is appropriate, meeting the needs of different users.

[0056] Embodiment 2 As Figure 2 shown, it is a schematic structural diagram of a map voice broadcast device 200 based on a large model in the embodiments of the present application. The device includes a server 210, a large model 220, and a client 230; The server 210 is used to generate the original broadcast copy for multiple broadcast positions in the current navigation, and input each original broadcast copy into the large model 220; The large model 220 is used to generate the large model broadcast copy for each broadcast position, and return each large model broadcast copy to the server 210; The server 210 is further used to obtain the scoring value of the historical navigation, generate a switching identifier according to the scoring value of the historical navigation, and send each large model broadcast copy, each original broadcast copy, and the switching identifier to the client 230; The client 230 is used to determine the actual broadcast copy for each broadcast position in the current navigation according to the switching identifier, record the corresponding actual broadcast information, calculate the actual broadcast score for each broadcast position in the current navigation according to each actual broadcast information, and simulate and calculate the large model broadcast score for each broadcast position in the current navigation, and send the actual broadcast score and the large model broadcast score for each broadcast position in the current navigation to the server 210; The server 210 is further used to calculate the overall broadcast score of the current navigation according to the actual broadcast score and the large model broadcast score for each broadcast position in the current navigation.

[0057] The map voice broadcast device based on a large model provided by the embodiments of the present application can implement each process of the map voice broadcast method based on a large model corresponding to Embodiment 1, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0058] The map voice broadcast device based on the large model provided by the embodiment of the present application trains the large model by collecting information such as the difference between the actual broadcast position and the expected broadcast position, the road grade, speed, and turning model of the broadcast position multiple times, generates a broadcast copy that conforms to the road conditions at a certain place, provides a corresponding discrimination algorithm to switch between the generated broadcast copy and the original broadcast copy, and also provides relevant calculation formulas to verify the effect of applying the broadcast copy of the large model to actual broadcasts. It not only improves the accuracy, timeliness, and personalization level of broadcasts, enhances the flexibility and adaptability of the device, reduces the development cost, but also can dynamically adjust the broadcast strategy according to the actual road conditions and driving habits to ensure that the broadcast content is accurate and the broadcast timing is appropriate, meeting the needs of different users.

[0059] Embodiment 3 The embodiment of the present application also provides a computer device. Specifically, please refer to Figure 3 , Figure 3 which is the basic structure block diagram of the computer device in this embodiment.

[0060] The computer device 3 includes a memory 31, a processor 32, and a network interface 33 that are communicatively connected to each other through a device bus. It should be noted that only the computer device 3 with a memory 31, a processor 32, and a network interface 33 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented. Among them, those skilled in the art of the present technology can understand that a computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0061] The computer device can be a desktop computer, a notebook, a palm computer, a cloud server, and other computing devices. The computer device can perform human-computer interaction with users through means such as a keyboard, a mouse, a remote control, a touchpad, or a voice control device.

[0062] The memory 31 includes at least one type of readable storage medium, which includes flash memory, hard disk, multimedia card, card-type memory (such as SD or D-slot compatibility test memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 31 may be an internal storage unit of the computer device 3, such as the hard disk or memory of the computer device 3. In other embodiments, the memory 31 may also be an external storage device of the computer device 3, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. equipped on the computer device 3. Of course, the memory 31 may also include both the internal storage unit and the external storage device of the computer device 3. In this embodiment, the memory 31 is generally used to store the operating device and various application software installed on the computer device 3, such as computer-readable instructions for the slot compatibility test method. In addition, the memory 31 may also be used to temporarily store various types of data that have been output or will be output.

[0063] In some embodiments, the processor 32 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other map voice broadcast chips based on large models. The processor 32 is generally used to control the overall operation of the computer device 3. In this embodiment, the processor 32 is used to run the computer-readable instructions stored in the memory 31 or process data, such as running the computer-readable instructions for the slot compatibility test method.

[0064] The network interface 33 may include a wireless network interface or a wired network interface, and this network interface 33 is generally used to establish a communication connection between the computer device 3 and other electronic devices.

[0065] The computer device provided in this embodiment can execute the above-mentioned map voice broadcast method based on large models. Here, the map voice broadcast method based on large models may be the map voice broadcast methods based on large models in the above various embodiments.

[0066] Embodiment 4 This embodiment also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the map voice broadcast method based on large models in the embodiment are implemented.

[0067] In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the computer-readable storage medium may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk equipped on the computer device, a Smart Media Card (SMC for short), a Secure Digital (SD for short) card, a Flash Card, etc. Of course, the computer-readable storage medium may also include both the internal storage unit and the external storage device of the computer device. In this embodiment, the computer-readable storage medium is generally used to store the operating device and various application software installed on the computer device. In addition, the computer-readable storage medium can also be used to temporarily store various data that have been output or will be output.

[0068] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and structure diagrams in the drawings show the possible architectures, functions, and operations of devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in an alternative implementation, the functions marked in the blocks may occur in a different order from that marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the structure diagram and / or flowchart, as well as the combination of blocks in the structure diagram and / or flowchart, can be implemented by a dedicated hardware-based device for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0069] In addition, each functional module or unit in various embodiments of the present invention may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0070] When the above-mentioned functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium may be a non-volatile storage medium or a volatile storage medium. For example, the storage medium may be: a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc, etc., which are various media that can store program codes.

[0071] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention.

Claims

1. A map voice announcement method based on a large model, characterized in that, The method includes: The server generates the original broadcast copywriting for multiple broadcast positions in the current navigation, and inputs each of the original broadcast copywritings into the large model; The large model generates the large model broadcast copywriting for each of the broadcast positions, and returns each of the large model broadcast copywritings to the server; The server obtains the scoring value of the historical navigation, generates a switching identifier according to the scoring value of the historical navigation, and sends each of the large model broadcast copywritings, each of the original broadcast copywritings, and the switching identifier to the client; The client determines the actual broadcast copywriting for each of the broadcast positions in the current navigation according to the switching identifier, records the corresponding actual broadcast information, calculates the actual broadcast score for each of the broadcast positions in the current navigation according to each of the actual broadcast information, and simulates and calculates the large model broadcast score for each of the broadcast positions in the current navigation, and sends the actual broadcast score and the large model broadcast score for each of the broadcast positions in the current navigation to the server; The server calculates the overall broadcast score of the current navigation according to the actual broadcast score and the large model broadcast score for each of the broadcast positions in the current navigation.

2. The method for map voice broadcast based on a large model according to claim 1, wherein, The switching identifier includes a first switching identifier and a second switching identifier. The obtaining the scoring value of the historical navigation and generating a switching identifier according to the scoring value of the historical navigation includes: Obtaining the scoring values of the previous two historical navigations, calculating the change value of the scoring values of the previous two historical navigations, and determining whether the change value is less than a preset convergence threshold; If the change value is less than the preset convergence threshold, then using each of the large model broadcast copywritings as the actual broadcast copywriting for the corresponding broadcast position in the current navigation, and generating the first switching identifier; If the change value is greater than or equal to the preset convergence threshold, then using each of the original broadcast copywritings as the actual broadcast copywriting for the corresponding broadcast position in the current navigation, and generating the second switching identifier.

3. The method for map voice broadcast based on a large model according to claim 2, wherein The determining the actual broadcast copywriting for each of the broadcast positions in the current navigation according to the switching identifier and recording the corresponding actual broadcast information includes: Determining the actual broadcast copywriting for each of the broadcast positions in the current navigation according to the first switching identifier or the second switching identifier; Recording the actual broadcast information when each of the broadcast positions in the current navigation broadcasts the corresponding actual broadcast copywriting, including the actual broadcast position, the best broadcast position, the broadcast upper limit position, the broadcast lower limit position, and the actual broadcast completion position.

4. The method for map voice broadcast based on large models according to claim 3, wherein, The calculating the actual broadcast score for each of the broadcast positions in the current navigation according to each of the actual broadcast information includes: Using a preset broadcast score calculation formula, and calculating the actual broadcast score for each of the broadcast positions in the current navigation according to the actual broadcast position, the best broadcast position, the broadcast upper limit position, the broadcast lower limit position, and the actual broadcast completion position in each of the actual broadcast information; Wherein, the preset broadcast score calculation formula is: Wherein, is the actual broadcast score at the i-th broadcast position, is the actual broadcast position at the i-th broadcast position, is the optimal broadcast position at the i-th broadcast position, is the upper limit position of the broadcast at the i-th broadcast position, is the lower limit position of the broadcast at the i-th broadcast position, is the actual broadcast completion position at the i-th broadcast position.

5. The method for map voice broadcast based on a large model according to claim 1, wherein The calculating the overall broadcast score of the current navigation according to the actual broadcast score and the large model broadcast score for each of the broadcast positions in the current navigation includes: Calculate the sum of the large model broadcast scores for all the said broadcast positions, calculate the sum of the actual broadcast scores for all the said broadcast positions, and obtain the total number of the said broadcast positions; Using a preset overall broadcast score calculation formula, calculate the overall broadcast score for this navigation based on the sum of the large model broadcast scores for all the said broadcast positions, the sum of the actual broadcast scores for all the said broadcast positions, and the total number of the said broadcast positions; Wherein, the preset overall broadcast score calculation formula is: Wherein, is the overall broadcast score of this navigation, is the total number of the said broadcast positions, is the large model broadcast score of the i-th broadcast position, is the sum of the large model broadcast scores of all the said broadcast positions, is the actual broadcast score of the i-th broadcast position, is the sum of the actual broadcast scores of all the said broadcast positions.

6. The method for map voice broadcast based on a large model according to claim 1, wherein The method further includes: The client uploads the actual broadcast copywriting, actual broadcast scores, and large model broadcast scores for each of the said broadcast positions in this navigation to the server; The server inputs the actual broadcast copywriting, actual broadcast scores, and large model broadcast scores for each of the said broadcast positions in this navigation into the large model; The large model corrects the corresponding actual broadcast copywriting according to the actual broadcast scores and large model broadcast scores for each of the said broadcast positions in this navigation to obtain multiple optimized broadcast copywritings, and returns each of the optimized broadcast copywritings to the server; The server simulates and calculates the optimized broadcast scores for each of the optimized broadcast copywritings, and feeds back the optimal broadcast copywriting with the highest optimized broadcast score to the large model; The large model is trained according to the optimal broadcast copywriting to obtain an optimized large model.

7. The method for map voice broadcast based on a large model according to claim 6, wherein, The correcting the corresponding actual broadcast copywriting according to the actual broadcast scores and large model broadcast scores for each of the said broadcast positions in this navigation to obtain multiple optimized broadcast copywritings includes: Judge whether the actual broadcast score for each of the said broadcast positions in this navigation is greater than a preset score threshold, and judge whether the large model broadcast score for each of the said broadcast positions in this navigation is greater than the preset score threshold; If the actual broadcast score for each of the said broadcast positions in this navigation is greater than the preset score threshold, and / or the large model broadcast score for each of the said broadcast positions in this navigation is greater than the preset score threshold, then correct each of the actual broadcast copywriting using a preset correction rule to obtain multiple of the optimized broadcast copywritings.

8. A map voice broadcast device based on a large model, characterized in that, The device includes a server, a large model, and a client; The server is used to generate the original broadcast copywriting for multiple broadcast positions in this navigation, and input each of the original broadcast copywriting into the large model; The large model is used to generate the large model broadcast copywriting for each of the said broadcast positions, and return each of the large model broadcast copywriting to the server; The server is further used to obtain the scoring value of the historical navigation, generate a switching identifier according to the scoring value of the historical navigation, and send each of the large model broadcast copywriting, each of the original broadcast copywriting, and the switching identifier to the client; The client is used to determine the actual broadcast copywriting for each of the said broadcast positions in this navigation according to the switching identifier, record the corresponding actual broadcast information, calculate the actual broadcast scores for each of the said broadcast positions in this navigation according to each of the actual broadcast information, and simulate and calculate the large model broadcast scores for each of the said broadcast positions in this navigation, and send the actual broadcast scores and large model broadcast scores for each of the said broadcast positions in this navigation to the server; The server is further configured to calculate the overall broadcast score of this navigation based on the actual broadcast scores and the large model broadcast scores of each of the broadcast positions in this navigation.

9. A computer device, characterized in that, It includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the large model-based map voice broadcast method described in any one of claims 1-7 are implemented.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the large model-based map voice broadcast method described in any one of claims 1-7 are implemented.

Images