Method and device for generating navigation broadcast text, electronic equipment and storage medium
By obtaining navigation elements, character information and navigation scene information, and using rewriting templates and rewriting models to generate multiple broadcast texts, solving the problem of rigid content of existing navigation broadcasts and improving the flexibility and diversity of user experience and broadcast content.
Patent Information
- Application Number
- CN202411111724.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-11-26
AI Technical Summary
When the existing navigation broadcasts realize the anthropomorphic function, the broadcast content is rigid, the user experience is poor, and there is lack of flexibility and richness.
By obtaining navigation elements, character information and navigation scene information, the two paths based on the rewrite template and the rewrite model are used for rewriting processing, multiple broadcast texts are generated, and navigation broadcast texts are selected from them, improving the flexibility and diversity of broadcast content. .
It realizes the richness and diversity of navigation broadcast content, improves user experience, and avoids the monotonicity and auditory fatigue of anthropomorphic style.
Smart Images

Figure CN119022913A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of navigation technology, and in particular to a method, device, electronic device and storage medium for generating navigation broadcast text. Background Art
[0002] Navigation broadcast is an indispensable part of the navigation system. It conveys navigation information to users through voice, helping users reach their destinations accurately and efficiently. In order to make navigation broadcast more personalized and in line with user preferences, some products have added anthropomorphic services. Users can choose the role information provided by the service, and the navigation broadcast content simulates the corresponding role characteristics.
[0003] However, most of the current navigation broadcasts only simulate the character's voice when realizing the anthropomorphic function. The broadcast content is rigid and the user experience needs to be improved urgently. Summary of the invention
[0004] In view of this, the present application provides a method, device, electronic device and storage medium for generating navigation broadcast text, so as to improve the flexibility of navigation broadcast content and enhance user experience.
[0005] This application provides the following solutions:
[0006] In a first aspect, a method for generating a navigation broadcast text is provided, the method comprising:
[0007] Get the navigation elements, character information and navigation scene information corresponding to the prompt point;
[0008] Performing a first rewriting process using the navigation element, the character setting information, and the navigation scene information to obtain at least one first broadcast text, wherein the first rewriting process is implemented based on a rewriting template;
[0009] Performing a second rewriting process using the navigation element, the character setting information, and the navigation scene information to obtain at least one second broadcast text, wherein the second rewriting process is implemented based on the rewriting model;
[0010] A navigation broadcast text is selected from the at least one first broadcast text and the at least one second broadcast text.
[0011] In a second aspect, a navigation broadcast method is provided, the method comprising:
[0012] Obtaining a navigation broadcast voice corresponding to a navigation broadcast text, wherein the navigation broadcast text is generated by the server using the method described in any one of the first aspects above;
[0013] Play the navigation announcement voice.
[0014] In a third aspect, a device for generating a navigation broadcast text is provided, the device comprising:
[0015] An information acquisition unit, configured to acquire navigation elements, character information, and navigation scene information corresponding to the prompt point;
[0016] A first rewriting unit is configured to perform a first rewriting process using the navigation element, the character setting information and the navigation scene information to obtain at least one first broadcast text, wherein the first rewriting process is implemented based on a rewriting template;
[0017] A second rewriting unit is configured to perform a second rewriting process using the navigation element, the character setting information and the navigation scene information to obtain at least one second broadcast text, wherein the second rewriting process is implemented based on the rewriting model;
[0018] The text selection unit is configured to select a navigation broadcast text from the at least one first broadcast text and the at least one second broadcast text.
[0019] According to a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the steps of the method described in any one of the first and second aspects are implemented.
[0020] According to a fifth aspect, there is provided an electronic device, comprising:
[0021] one or more processors; and
[0022] A memory associated with the one or more processors, the memory being used to store program instructions, wherein the program instructions, when read and executed by the one or more processors, execute the steps of the method described in any one of the first aspects above.
[0023] According to a fifth aspect, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of the method described in any one of the first and second aspects above.
[0024] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0025] The present application adopts two paths based on the navigation elements, character information and navigation scene information corresponding to the prompt points, namely, a first rewriting process based on a rewriting template and a second rewriting process based on a rewriting model, to obtain multiple broadcast texts, namely, at least one first broadcast text and at least one second broadcast text, and select a navigation broadcast text from them, so that the navigation broadcast content is more flexible and rich, and the user experience is improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0027] Figure 1 is a system architecture diagram applicable to the embodiments of the present application;
[0028] Figure 2 A flowchart of a method for generating navigation broadcast text provided in an embodiment of the present application;
[0029] Figure 3 A schematic diagram of a method for implementing a first rewriting process based on a rewriting template provided in an embodiment of the present application;
[0030] Figure 4 A schematic diagram of a method for implementing a second rewriting process based on a rewriting model provided in an embodiment of the present application;
[0031] Figure 5 A schematic block diagram of a device for generating navigation broadcast text provided in an embodiment of the present application;
[0032] Figure 6 A schematic block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0033] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.
[0034] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.
[0035] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.
[0036] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.
[0037] In view of this, the present application provides a new idea. In order to facilitate the understanding of the present application, the system architecture on which the present application is based is first described. Figure 1 An exemplary system architecture to which the embodiments of the present application can be applied is shown. Figure 1 As shown in , the system architecture may include: a user terminal and a server.
[0038] The server provides navigation broadcast services to users. When users use navigation during driving, they can obtain the navigation broadcast voice provided by the server through the user terminal.
[0039] The user terminal may include, but is not limited to, smart mobile terminals, wearable devices, etc. Smart mobile devices may include, for example, mobile phones, tablet computers, PDAs (Personal Digital Assistants), Internet car terminals, etc. Wearable devices may include, for example, smart watches, smart glasses, smart bracelets, VR (Virtual Reality) devices, AR (Augmented Reality), mixed reality devices (i.e., devices that can support both virtual reality and augmented reality), etc.
[0040] The server side may include a device for generating navigation broadcast text and a speech synthesis device. The device for generating navigation broadcast text may use the method provided in the embodiment of the present application to generate navigation broadcast text, and provide the navigation broadcast text to the speech synthesis device. The speech synthesis device may use a technology such as TTS (text-to-speech) to convert the navigation broadcast text into navigation broadcast voice, and send the navigation broadcast voice to the user terminal for broadcasting during the user's driving process.
[0041] The navigation text broadcast device and the speech synthesis device can be set up as independent servers, or can be set up on the same server or server group, or can be set up on independent or the same cloud server. Cloud server, also known as cloud computing server or cloud host, is a host product in the cloud computing service system to solve the defects of difficult management and weak service scalability in traditional physical host and virtual private server (VPS) services.
[0042] It should be understood that Figure 1 The number of navigation text broadcasting devices, speech synthesis devices and user terminals in the embodiment is only for illustration. According to the implementation requirements, there may be any number of navigation text broadcasting devices, speech synthesis devices and user terminals.
[0043] Figure 2 A flowchart of a method for generating navigation broadcast text provided in an embodiment of the present application, the method can be performed by Figure 1 The device for generating navigation broadcast text in the system shown is executed. Figure 2 As shown in , the method may include the following steps:
[0044] Step 201: Obtain navigation elements, character information and navigation scene information corresponding to the prompt point.
[0045] Step 203: Perform a first rewriting process using the navigation elements, the character information, and the navigation scene information to obtain at least one first broadcast text, wherein the first rewriting process is implemented based on a rewriting template.
[0046] Step 205: Perform a second rewriting process using the navigation elements, the character information, and the navigation scene information to obtain at least one second broadcast text, wherein the second rewriting process is implemented based on the rewriting model.
[0047] It should be noted that the above steps 203 and 205 can be executed in any order or in parallel. Figure 2 Take parallel execution as an example.
[0048] Step 207: Select a navigation broadcast text from the at least one first broadcast text and the at least one second broadcast text.
[0049] It can be seen from the above process that the present application adopts two paths based on navigation elements, character information and navigation scene information, namely, a first rewriting process based on a rewriting template and a second rewriting process based on a rewriting model, to obtain multiple broadcast texts, namely, at least one first broadcast text and at least one second broadcast text, and select a navigation broadcast text from them, which makes the navigation broadcast content more flexible and improves the user experience.
[0050] The following is a detailed description of each step in the above process and the effects that can be further produced in conjunction with the embodiments. It should be noted that the limitations such as "first" and "second" involved in the present disclosure do not have limitations in terms of size, order, and quantity, and are only used to distinguish them in name. For example, "first rewriting process" and "second rewriting process" are only used to distinguish the two rewriting processes in name. For another example, "first broadcast text" and "second broadcast text" are only used to distinguish the two broadcast texts in name.
[0051] First, the above step 201, namely "obtaining the navigation element, character setting information and navigation scene information corresponding to the prompt point" is described in detail in conjunction with the embodiment.
[0052] Usually, navigation broadcast occurs in the process of providing navigation services to users. The so-called navigation service refers to a service that provides path guidance to users based on the user's starting location and target location. During the navigation service, navigation voice broadcast will be performed at each prompt point. The navigation voice is usually obtained through speech synthesis processing based on the navigation broadcast text. In other words, the navigation broadcast text needs to be generated for each prompt point.
[0053] The so-called prompt points, also known as induction points, refer to the information prompt points set during the navigation process, which are used to broadcast navigation at each prompt point. The prompt points can be some key locations (such as bridge entrances and exits, tunnel entrances and exits, intersections, etc.), or some environmental information (such as locations with congested traffic, gas stations, rest areas, etc.), and navigation actions (such as turning, speed limit, etc.).
[0054] In an embodiment of the present application, after determining the navigation path, the method for generating navigation broadcast text provided in the embodiment of the present application can be executed for each prompt point in the navigation path, thereby obtaining the navigation broadcast text corresponding to each prompt point. After the navigation broadcast text is determined for each prompt point in advance, the navigation voice corresponding to the navigation broadcast text is sent to the user terminal after the user approaches or arrives at the prompt point. It is also possible to determine the navigation broadcast text corresponding to the prompt point after the user approaches or arrives at the prompt point, and send the navigation voice corresponding to the navigation broadcast text to the user terminal.
[0055] Navigation elements are usually determined based on the navigation route, reflecting the location information, environmental information or navigation actions of the prompt points, and are the core content of the navigation broadcast text. For example, "gas station ahead", "turn right", "U-turn after 100 meters", "road congestion ahead", etc.
[0056] The navigation scene is determined based on at least one of the user's location and the environment. The navigation scene can be divided according to different standards and can adopt different division granularities. For example, the navigation scene can be such as a guidance scene (such as guidance for turning, guidance for changing lanes, etc.), a confirmation scene (usually corresponding to a scene that is executed immediately, such as a right turn, a U-turn, etc.), a reminder scene (such as an electronic eye, a speed limit), etc.
[0057] The character setting information is determined based on the role information selected by the user. It has been mentioned in the background technology that in order to make the navigation broadcast more personalized and close to user preferences, anthropomorphic services have been added to some products, and users can choose the role information provided by the service. These characters can be specific people (such as some well-known stars), virtual characters (such as some animation characters, movie characters, etc.), or character labels with specific characteristics (such as sweet girls, gentle and warm men, funny girls, middle school boys, etc.). Different characters have different characteristics and correspond to different portraits, such as age, occupation, catchphrases, interests, styles, etc.
[0058] The above navigation elements, navigation scene information and character information can be obtained by the device generating the navigation broadcast text requesting other devices on the server side, for example, obtaining navigation elements and navigation scene information from the navigation planning device, and obtaining character information corresponding to the role information selected by the user from the personalized service device. This application does not limit the method for obtaining navigation elements, navigation scene information and character information.
[0059] The above step 203, namely "performing a first rewriting process using the navigation elements, character information and navigation scene information to obtain at least one first broadcast text" is described in detail below in conjunction with an embodiment.
[0060] The first rewriting process involved in this step is implemented based on a rewriting template. A rewriting template can be constructed in advance for the character information and navigation scene information. After obtaining the navigation element, character information and navigation scene information corresponding to the prompt point, at least one rewriting template corresponding to the character information and the navigation scene information can be first determined, and then the navigation element and the at least one rewriting template can be used to obtain at least one first broadcast text.
[0061] For the same combination of character information and navigation scene information, the corresponding rewriting template can be one or more. The rewriting templates corresponding to the character information and navigation scene information can be stored in the first configuration table in advance. The first configuration table can be stored on the server side, such as the cloud, for query. It can be considered that the first configuration table stores the correspondence between each character information, navigation scene information and rewriting template. Here is an example to illustrate the form that the first configuration table may take, as shown in Table 1 below. In the first configuration table, one column stores character information, one column stores navigation scene information, and another column stores rewriting templates. The same row represents the correspondence between character information, navigation scene information and rewriting templates. For example, the combination of character a and scene 1 corresponds to rewriting template A; the combination of character a and scene 2 corresponds to rewriting template B and rewriting template C; and so on.
[0062] Table 1
[0063]
[0064]
[0065] like Figure 3 As shown in , the first configuration table is queried using the character information and navigation scene information, and the rewriting template found is used as a candidate rewriting template corresponding to the character information and navigation scene information obtained in step 201. Among them, if the character information and the navigation scene information in the first configuration table correspond to one rewriting template, then one candidate rewriting template is obtained for the current prompt point; if the character information and the navigation scene information in the first configuration table correspond to multiple rewriting templates, then multiple candidate rewriting templates are obtained for the current prompt point. Still taking Table 1 as an example, assuming that the character information and the navigation scene information of the current prompt point are character a and scene 3 respectively, then by querying the first configuration table shown in Table 1, rewriting template D can be obtained as a candidate rewriting template. Assuming that the character information and the navigation scene information of the current prompt point are character a and scene 2 respectively, then by querying the first configuration table shown in Table 1, rewriting template B and rewriting template C can be obtained as candidate rewriting templates.
[0066] By querying the first configuration table, at least one rewriting template corresponding to the character information and navigation scene information can be quickly located, and the first broadcast text is obtained by using the rewriting template and navigation elements corresponding to the character information and navigation scene information. On the one hand, the rewriting method based on the rewriting template helps to unify the anthropomorphic style, and on the other hand, it is highly efficient and can meet the real-time requirements of online navigation broadcasts.
[0067] When constructing the first configuration table, it may be a rewriting template manually set for each character setting information and navigation scene information, or it may be a rewriting template obtained by mining the historical broadcast text corresponding to the historical character setting information and navigation scene information.
[0068] The rewrite template may include navigation element slots, fixed words, and hook word slots. For example, if the character is a sweet girl and the navigation scenario is a confirmation scenario, one of the corresponding rewrite templates may be:
[0069] "[Navigation element] enter the internal path [hook word]"
[0070] Among them, "[Navigation element]" is a navigation element slot, which is used to assign a value to the slot using a navigation element. "Enter the internal route" is a fixed phrase. "[Hook word]" is a hook word slot, which is used to assign a value to the slot using a hook word.
[0071] In addition to the above-mentioned methods, the rewriting template may also adopt other methods, such as including navigation element slots, fixed words and fixed hook words. However, in order to further ensure the flexibility of the navigation broadcast text under the anthropomorphic service, the former implementation method is preferred, that is, the rewriting template includes navigation element slots, fixed words and hook word slots.
[0072] After obtaining the candidate rewriting templates, at least one rewriting template can be selected from the candidate rewriting templates. The selection can be made randomly or according to a preset order or rule. However, as one of the preferred implementations, Figure 3 As shown in , at least one rewriting template can be selected from the obtained candidate rewriting templates according to a preset first probability distribution.
[0073] The first probability distribution includes the selection probability corresponding to each candidate rewriting template, that is, each candidate rewriting template has its own probability of being selected, and the selection probabilities corresponding to different candidate rewriting templates can be the same or different. The selection probability corresponding to each candidate rewriting template is preset and can be set according to experience values, experimental values, etc.
[0074] For example, assuming that the character information and navigation scene information of the current prompt point are character a and scene 2, then by querying the first configuration table shown in Table 1, rewriting template B and rewriting template C can be obtained as candidate rewriting templates. If the selection probabilities corresponding to rewriting template B and rewriting template C are 60% and 40% respectively, it means that rewriting template B and rewriting template C have a chance to be selected every time, but the probabilities of being selected are different: the probability of rewriting template B being selected each time is 60%, and the probability of rewriting template C being selected each time is 40%. If 10 selections are made, rewriting template B will be selected 6 times, and rewriting template C will be selected 4 times. The principle of selecting according to probability distribution in subsequent embodiments is similar and will not be repeated.
[0075] By selecting the rewriting template based on the first probability distribution, the navigation broadcast text finally generated can be made richer and more diverse, thereby improving the user experience.
[0076] After selecting at least one rewriting template, when obtaining at least one first broadcast text using the navigation element and the at least one rewriting template, Figure 3 As shown in , the navigation element slot in the rewriting template can be assigned a value by using the navigation element, for example, filling the navigation element into the navigation element slot.
[0077] If the rewriting template contains a hook word slot, the character information obtained in step 201 can be further used to query the second configuration table to obtain at least one hook word corresponding to the character information. The second configuration table includes the correspondence between each character information and the hook word. When constructing the second configuration table, it can be a hook word manually set for each character information, or it can be obtained by mining the historical words corresponding to the character information, such as extracting commonly used hook words from various historical words said by a celebrity.
[0078] Hook words are one of the important elements of personification. They are not specific navigation content, but some auxiliary words. These auxiliary words usually use the expression and tone corresponding to the character information. For example, if the character is a sweet girl, the corresponding hook words can be such as "Remember", "Don't miss it", "Brother", etc.
[0079] After obtaining at least one attachment word corresponding to the character setting information, a attachment word may be selected therefrom to assign a value to the attachment word slot in the rewriting template, for example, the attachment word may be filled into the attachment word slot.
[0080] Continuing with the previous example, the template is rewritten as "
Navigation element
Hook word
[0081] When selecting the hook words, they can be selected randomly or according to a preset order or rule. However, as one of the preferred implementations, Figure 3 As shown in , a hanging word can be selected from at least one hanging word obtained by querying the second configuration table according to a preset second probability distribution.
[0082] Among them, the second probability distribution includes the selection probability corresponding to each attachment word, that is, each attachment word has its own probability. This method of flexibly selecting attachment words based on probability further improves the flexibility and diversity of the first broadcast text, thereby improving the flexibility and diversity of the navigation broadcast text, and to a certain extent improves the controllability of the attachment words. The selection probabilities corresponding to different attachment words can be the same or different. The selection probability corresponding to each attachment word is preset and can be set according to experience values, experimental values, etc.
[0083] Since there may be multiple candidate rewriting templates obtained by query, there may be multiple rewriting templates selected from the candidate rewriting templates, and there may be multiple attachment words obtained, multiple first broadcast texts may be obtained through combination.
[0084] The rewriting template selected from the candidate rewriting templates may be one, and the obtained hook word may also be one or selected from them, so it is also possible to obtain a first broadcast text.
[0085] In addition, as mentioned above, the hook words reflect the tone to a certain extent, and the tone levels of different hook words are different. Some are particularly strong, such as "Brother", "Don't miss it", etc. Some are weaker, such as "Ah", "Yeah". If the navigation broadcast texts used in the navigation process are all strong in tone, such as frequent "Brother, turn right 100 meters ahead", "Exit ahead, don't miss it", "Brother, congestion ahead", etc., users are likely to feel boredom, auditory fatigue, etc. Therefore, as one of the more preferred implementation methods, in the second probability distribution, the tone level corresponding to each hook word can be negatively correlated with the selection probability. The stronger the tone level of the hook word, the smaller the probability of being selected, and the weaker the tone level of the hook word, the greater the probability of being selected, thereby avoiding user boredom and auditory fatigue and improving user experience.
[0086] The above step 205, i.e. "performing a second rewriting process using navigation elements, character information and navigation scenario information to obtain at least one second broadcast text, wherein the second rewriting process is implemented based on a rewriting model", is described in detail below in conjunction with an embodiment.
[0087] This step is actually based on the anthropomorphism of the broadcast text realized by the rewriting model. Since the corresponding broadcast words can usually be determined based on the navigation elements, for example, for the navigation element "turn right", the corresponding broadcast words can be such as "turn right ahead". Or, in addition to the navigation elements, the corresponding broadcast words can also be determined in combination with the navigation scene information. For example, for the navigation element "turn right", the determined navigation scene is "turn and reach the destination", then the corresponding broadcast words can be such as "turn right, then reach the destination", and can also be combined with other information, which are not listed here one by one. Therefore, in this step, the broadcast words corresponding to the navigation element obtained in step 201 can be first determined, and then at least one second broadcast text corresponding to the broadcast words can be obtained. The second broadcast text is obtained by rewriting the broadcast words by the rewriting model using the human information and navigation scene information obtained in step 201. This method makes full use of the rewriting model's understanding and rewriting ability of natural language, and the rewriting process of the rewriting model is relatively freer, so it can produce more diverse second broadcast texts.
[0088] When obtaining at least one second broadcast text corresponding to the broadcast speech, the following two methods may be used but are not limited to:
[0089] The first method is to provide the broadcast words, character information and navigation scene information corresponding to the navigation element obtained in step 201 to a rewriting model, obtain at least one broadcast text obtained by the rewriting model rewriting the broadcast words using the character information and navigation scene information, and obtain a second broadcast text using the at least one broadcast text.
[0090] This method is actually a process of rewriting in real time using a rewriting model online to obtain a second broadcast text. In an embodiment of the present application, the rewriting model can adopt an ordinary pre-trained language model or a large language model to achieve the rewriting of the broadcast speech corresponding to the navigation element, and the constraint information during rewriting is determined by the person setting information and the navigation scene information. Among them, a large language model is preferably used to achieve it. The large language model, i.e., LLM (Large Language Model), refers to a deep learning model trained with a large amount of text data, which can generate natural language text or understand the meaning of language text. It is characterized by a large scale and a huge amount of parameters (usually reaching more than 10 billion levels), and is usually based on a deep learning architecture such as a Transformer architecture. The difference between LLM and an ordinary pre-trained language model is the parameter scale. When the parameter scale exceeds a certain level, the model achieves a significant performance improvement and shows the ability that does not exist in a small model, such as in-context learning ability, which can learn complex patterns in language and perform a wide range of tasks, including text summarization, translation, sentiment analysis, multi-round dialogue, etc. Therefore, in order to distinguish it from traditional pre-trained language models, models with parameter scale exceeding a certain level are called LLMs. In general, language models with parameter scale exceeding 10 billion based on deep learning architecture can be considered as large language models. Common LLMs include: GTP-3 (Generative Pre-trained Transformer 3), T5 (Text-to-Text Transfer Transformer), GTP-4, PaLM (a large language model proposed by Google), LLaMA (Large Language Model Meta AI, a large language model released by Meta AI), etc.
[0091] At present, the large language model has a wealth of knowledge in the process of pre-training, so it has strong language understanding and reasoning capabilities, and can complete the prediction task specified by the prompt instruction given a prompt instruction (Prompt). In this embodiment, the prompt instruction can be constructed using the broadcasting speech, character information and navigation scene information, and then input into the large language model to rewrite the broadcasting speech to obtain at least one broadcast text. Among them, the prompt instruction can be constructed based on a manually set prompt template, which expresses the format of the template and can include broadcasting speech slots, character slots, navigation scene slots and fixed speech. After assigning values to the broadcasting speech slots, character slots and navigation scene slots in the prompt template using the broadcasting speech, character information and navigation scene information, the prompt instruction is obtained.
[0092] The broadcast text obtained by rewriting the template can be one or more. In the embodiment of the present application, the broadcast text obtained by the rewriting model can be filtered to filter out the broadcast text that does not comply with laws and regulations, contains sensitive content, or contains obvious erroneous content, thereby obtaining a second broadcast text. The filtering process can be implemented based on pre-set filtering rules or dictionaries.
[0093] The second method: use the broadcast words, character information and navigation scene information corresponding to the navigation element obtained in step 201 to query the rewrite text library, obtain candidate broadcast texts corresponding to the broadcast words, character information and navigation scene information, and select at least one second broadcast text from the candidate broadcast texts, wherein the rewrite text library is obtained by rewriting the historical broadcast words in advance by the rewriting model using the character information and navigation scene information.
[0094] This method actually uses the rewriting model to pre-build a rewriting text library in the offline stage. During the navigation broadcast process, the rewriting text library is queried online to determine the candidate broadcast texts corresponding to the broadcast words, character information and navigation scene information, and then at least one second broadcast text is selected from the candidate broadcast texts.
[0095] When building a rewritten text library, Figure 4 As shown in , high-frequency historical navigation scene information and high-frequency historical broadcast words can be pre-screened from the historical broadcast logs. The so-called high frequency refers to the number of occurrences within a preset time length being greater than or equal to a preset number threshold. Then, based on the preset character information, the above-selected navigation scene information and historical broadcast words, various combinations are obtained (a combination includes character information, navigation scene information and broadcast words), and prompt instructions are constructed based on each combination, and the prompt instructions are provided to the rewriting model to obtain the broadcast text obtained by the rewriting model for each combination. The combination and the broadcast text are then stored in the rewriting text library in correspondence.
[0096] Furthermore, in the embodiment of the present application, the broadcast text obtained by the rewriting model can be filtered to filter out the broadcast text that does not comply with laws and regulations, contains sensitive content, or contains obvious erroneous content, and then the filtered broadcast text and its corresponding combination are stored in the rewriting text library. The filtering process can be implemented based on pre-set filtering rules or dictionaries, or can be implemented based on manual review.
[0097] The first method uses the rewriting model to rewrite in real time online, which saves more storage resources (for example, there is no need to pre-build and store the rewriting text library). The second method uses the rewriting model to build the rewriting text library in advance, and rewrites by directly querying the rewriting text library online. This method has a faster response speed and is easier to review and control the broadcast content.
[0098] Continue to see Figure 4 In the navigation broadcast process, the broadcast script, character information, and navigation scene information corresponding to the navigation element obtained in step 201 can be used to query the rewritten text library. If a matching broadcast text is found, the found broadcast text is used as a candidate broadcast text. The candidate broadcast text can be one or more. All the found candidate broadcast texts can be selected as the second broadcast text, or one or more can be selected from the candidate broadcast texts as the second broadcast text.
[0099] For example, assuming that step 201 obtains the navigation element "turn right", the character information "sweet girl", and the navigation scene "turn and reach the destination", the corresponding announcement words of the navigation element are "turn right, and then reach the destination". After querying the rewriting text library, the announcement text corresponding to the combination of "turn right, and then reach the destination", "sweet girl" and "turn and reach the destination" is "turn right, and then the destination is reached, you are so loyal to have a beginning and an end", and this announcement text is a candidate announcement text, which can be used as the second announcement text.
[0100] The above step 207, ie, "selecting a navigation broadcast text from the above at least one first broadcast text and the above at least one second broadcast text", is described in detail below in conjunction with an embodiment.
[0101] In step 203 and step 205 of the embodiment of the present application, two rewriting methods are respectively adopted, namely, rewriting based on a rewriting template and rewriting based on a rewriting model. The rewriting principles of the two methods are different. Therefore, the first broadcast text and the second broadcast text obtained respectively will also have certain differences in rewriting logic, anthropomorphic style, etc., and each has its own characteristics. For example, the first broadcast text obtained by the first rewriting process based on the rewriting template usually has a more unified anthropomorphic style, but it is easy to cause the rewritten content to be relatively monotonous. The second broadcast text obtained by the second rewriting process based on the rewriting model is more diverse, but the rewritten content is not easy to control and usually requires manual review. Therefore, in order to enrich the user experience and avoid fatigue caused by unified rewriting logic and anthropomorphic style, the embodiment of the present application combines the above two rewriting methods. In the actual navigation broadcast process, one of the at least one first broadcast text and the at least one second broadcast text is flexibly selected as the navigation broadcast text.
[0102] As one of the feasible ways, one may be randomly and flexibly selected from the at least one first broadcast text and the at least one second broadcast text as the navigation broadcast text.
[0103] As another feasible method, one may be selected as a navigation broadcast text from the at least one first broadcast text and the at least one second broadcast text according to a preset third probability distribution. The third probability distribution includes the selection probabilities corresponding to the first broadcast text and the second broadcast text, respectively, which may also be understood as the selection probabilities corresponding to the first rewriting processing method based on the rewriting template and the second rewriting processing method based on the rewriting model. In other words, the first rewriting processing method based on the rewriting template and the second rewriting processing method based on the rewriting model have their own probabilities of being selected, which may correspond to the same probability or different probabilities. The specific probability values may be preset according to empirical values, experimental values, etc.
[0104] The above method can realize the personalized service of navigation broadcast, and in the personalized service of navigation broadcast at each prompt point, the diversity of navigation broadcast text is improved, the user's auditory fatigue is reduced, and the user experience is improved.
[0105] Furthermore, since the user will obtain the navigation voice corresponding to the navigation broadcast text at each prompt point in the process of obtaining the navigation broadcast service, if the navigation broadcast text for each prompt point adopts the above-mentioned anthropomorphic service, it may also make the user feel bored. For example, there may be multiple consecutive prompt points such as "The speed limit ahead is 60, Xiao Xin, don't speed", "Brother, turn right in 100 meters ahead", "Brother, then the destination is here, you are so loyal to have a beginning and an end", etc. Continuous prompt points have strong anthropomorphic services that can easily make users feel auditory fatigue and boredom. Therefore, before executing the method provided in this application for the prompt point, it can be determined whether the current prompt point adopts the above-mentioned fourth probability distribution according to the preset fourth probability distribution. Figure 2 If yes, continue to execute Figure 2 Otherwise, a common navigation broadcast service, such as a non-anthropomorphic navigation broadcast service, may be used.
[0106] Among them, the fourth probability distribution includes each prompt point using Figure 2 That is, for each prompt point in the same navigation route, there is a corresponding anthropomorphic service probability. Whether to use the anthropomorphic service at the prompt point is determined according to the anthropomorphic service probability corresponding to the prompt point. Figure 2 The method for generating navigation announcement text is shown. The probabilities of anthropomorphic services corresponding to different prompt points may be the same or different, and may be preset according to experience values, experimental values, etc. Whether anthropomorphic services are used at each prompt point is controlled by the fourth probability distribution, which can further improve the user experience during the overall announcement process and reduce the auditory fatigue of the user caused by the continuous provision of anthropomorphic services at consecutive prompt points.
[0107] It can be seen from the above embodiments that the embodiment of the present application can control the degree of anthropomorphic service of the navigation broadcast from four dimensions, namely: in the first rewriting process based on the rewriting template, the rewriting template is selected from the candidate rewriting templates based on the first distribution probability; in the first rewriting process based on the rewriting template, the hook words filled in the rewriting template are selected based on the second distribution probability; based on the third probability distribution, the navigation broadcast text is selected from the first broadcast text obtained by the first rewriting process based on the rewriting template and the second broadcast text obtained by the second rewriting process based on the rewriting model; based on the fourth distribution probability, it is determined whether the anthropomorphic service provided by the embodiment of the present application is effective at each prompt point. Through the control of these four dimensions, it is possible to achieve a unified anthropomorphic style based on the person information in different navigation scenarios, generate more abundant and diverse navigation broadcast texts, and effectively avoid the user's auditory fatigue, thereby improving the user experience.
[0108] For a user terminal running on a user terminal, the navigation broadcast method executed by the user terminal may include: obtaining a navigation broadcast voice corresponding to a navigation broadcast text, wherein the navigation broadcast text is a navigation broadcast text generated by a server using Figure 2 The method flow shown generates; playing navigation voice.
[0109] The user terminal may be a client running on a user terminal, a small program, or a Web application running through a browser.
[0110] As one of the achievable ways, after obtaining the navigation broadcast text of the prompt point, the server can provide the navigation broadcast text of the prompt point to the speech synthesis device on the server side, and the speech synthesis device converts the navigation broadcast text into navigation broadcast voice and sends it to the user side for navigation voice broadcast. When providing the navigation broadcast text to the speech synthesis device, the persona information can be further provided to the speech synthesis device, and the speech synthesis device can determine the corresponding acoustic information such as tone, speech speed, timbre, etc. according to the persona information, and use the acoustic information to convert the navigation broadcast text into navigation broadcast voice.
[0111] As another achievable method, a speech synthesis device can be provided at the user end, that is, the user end has a speech synthesis function. After obtaining the navigation broadcast text of the prompt point, the server can send the navigation broadcast text of the prompt point to the user end. The user end converts the navigation broadcast text into navigation broadcast voice and then broadcasts it.
[0112] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0113] According to another embodiment, a device for generating navigation announcement text is provided. Figure 5 As shown, the device 500 may include: an information acquisition unit 501, a first rewriting unit 502, a second rewriting unit 503 and a text selection unit 504. The main functions of each component unit are as follows:
[0114] The information acquisition unit 501 is configured to acquire the navigation element, character information and navigation scene information corresponding to the prompt point.
[0115] The first rewriting unit 502 is configured to perform a first rewriting process using the navigation element, the character setting information and the navigation scene information to obtain at least one first broadcast text, wherein the first rewriting process is implemented based on the rewriting template.
[0116] The second rewriting unit 503 is configured to perform a second rewriting process using the navigation elements, the character information and the navigation scene information to obtain at least one second broadcast text, and the second rewriting process is implemented based on the rewriting model.
[0117] The text selection unit 504 is configured to select a navigation broadcast text from at least one first broadcast text and at least one second broadcast text.
[0118] Among them, the first rewriting unit 502 can be specifically configured as: using the first configuration table, determining at least one rewriting template corresponding to the character information and the navigation scene information, and obtaining at least one first broadcast text using the navigation element and at least one rewriting template; wherein the first configuration table includes the correspondence between the character information, the navigation information and the rewriting template.
[0119] As one of the feasible ways, when determining at least one rewriting template corresponding to the character information and navigation scene information, the first rewriting unit 502 can use the character information and navigation scene information obtained by the information acquisition unit 501 to query the first configuration table, and use the rewriting template obtained by the query as a candidate rewriting template; according to a preset first probability distribution, select at least one rewriting template from the candidate rewriting templates; wherein the first probability distribution includes the selection probability corresponding to each candidate rewriting template.
[0120] The rewriting template may include a navigation element slot, a fixed phrase, and a hook word slot. As one of the achievable methods, when the first rewriting unit 502 obtains at least one first broadcast text using the navigation element and at least one rewriting template, it may use the persona information to query the second configuration table to obtain at least one hook word corresponding to the persona information; use the navigation element to assign a value to the navigation element slot in at least one rewriting template, and use the at least one hook word to assign a value to the hook word slot in the at least one rewriting template to obtain at least one first broadcast text; wherein the second configuration table includes the correspondence between the persona information and the hook word.
[0121] As one of the more preferred implementation methods, when the first rewriting unit 502 uses at least one attachment word to assign a value to the attachment word slot in the above-mentioned at least one rewriting template, it can select an attachment word from at least one attachment word according to a preset second probability distribution to assign a value to the attachment word slot in at least one rewriting template; wherein the second probability distribution includes the selection probability corresponding to each attachment word.
[0122] As a more preferred implementation, in the second probability distribution, the tone level corresponding to each linking word is negatively correlated with the selection probability.
[0123] Among them, the second rewriting unit 503 can be specifically configured to: determine the broadcast words corresponding to the navigation element, obtain at least one second broadcast text corresponding to the broadcast words, and the second broadcast text is obtained by rewriting the broadcast words by the rewriting model using the character information and the navigation scene information.
[0124] As one of the feasible ways, when the second rewriting unit 503 obtains at least one second broadcast text corresponding to the broadcast words, it can provide the broadcast words, character information and navigation information to the rewriting model, obtain at least one broadcast text obtained by rewriting the broadcast words by the rewriting model using the character information and navigation information, and obtain the second broadcast text using the at least one broadcast text.
[0125] As another achievable method, when obtaining at least one second broadcast text corresponding to the broadcast words, the second rewriting unit 503 can use the broadcast words, character information and navigation scene information to query the rewriting text library, obtain candidate broadcast texts corresponding to the broadcast words, character information and navigation scene information, and select at least one second broadcast text from the candidate broadcast texts, wherein the rewriting text library is obtained by rewriting the historical broadcast words in advance by the rewriting model using the character information and navigation scene information.
[0126] As one of the achievable methods, the text selection unit 504 may be specifically configured to: determine whether the current prompt point adopts the method of generating navigation broadcast text based on a preset fourth probability distribution, and if so, continue to execute the method of generating navigation broadcast text; wherein the fourth probability distribution includes the probability of each prompt point adopting the method of generating navigation broadcast text.
[0127] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.
[0128] 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, stored data, displayed data, 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 relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.
[0129] In addition, an embodiment of the present application further provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of any one of the methods in the aforementioned method embodiments are implemented.
[0130] And an electronic device, comprising:
[0131] one or more processors; and
[0132] A memory associated with the one or more processors, the memory being used to store program instructions, wherein the program instructions, when read and executed by the one or more processors, execute the steps of the method described in any one of the aforementioned method embodiments.
[0133] The present application also provides a computer program product, including a computer program, which implements the steps of any one of the methods in the aforementioned method embodiments when executed by a processor.
[0134] in, Figure 6 The architecture of the electronic device is shown as an example, which may include a processor 610, a video display adapter 611, a disk drive 612, an input / output interface 613, a network interface 614, and a memory 620. The processor 610, the video display adapter 611, the disk drive 612, the input / output interface 613, the network interface 614, and the memory 620 may be communicatively connected via a communication bus 630.
[0135] Among them, the processor 610 can be implemented by a general-purpose CPU, a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., to execute relevant programs to implement the technical solutions provided in this application.
[0136] The memory 620 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 620 can store an operating system 621 for controlling the operation of the electronic device 600, and a basic input and output system (BIOS) 622 for controlling the low-level operation of the electronic device 600. In addition, a web browser 623, a data storage management system 624, and a device 625 for generating navigation broadcast text, etc. can also be stored. The above-mentioned device 625 for generating navigation broadcast text can be an application program that specifically implements the operations of the aforementioned steps in the embodiment of the present application. In short, when the technical solution provided by the present application is implemented by software or firmware, the relevant program code is stored in the memory 620 and is called and executed by the processor 610.
[0137] The input / output interface 613 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.
[0138] The network interface 614 is used to connect to a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).
[0139] The bus 630 comprises a pathway for transmitting information between the various components of the device (eg, the processor 610, the video display adapter 611, the disk drive 612, the input / output interface 613, the network interface 614, and the memory 620).
[0140] It should be noted that, although the above device only shows a processor 610, a video display adapter 611, a disk drive 612, an input / output interface 613, a network interface 614, a memory 620, a bus 630, etc., in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include components necessary for implementing the solution of the present application, and does not necessarily include all the components shown in the figure.
[0141] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly contributed to the prior art in the form of a computer program product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.
[0142] The technical solution provided by the present application is described in detail above. The principle and implementation method of the present application are described in detail using specific examples. The description of the above embodiments is only used to help understand the method and core idea of the present application. At the same time, for those skilled in the art, according to the idea of the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present application.
Claims
1. A method for generating navigation broadcast text, characterized in that: The method comprises: Get the navigation elements, character information and navigation scene information corresponding to the prompt point; Performing a first rewriting process using the navigation element, the character setting information, and the navigation scene information to obtain at least one first broadcast text, wherein the first rewriting process is implemented based on a rewriting template; Performing a second rewriting process using the navigation element, the character setting information, and the navigation scene information to obtain at least one second broadcast text, wherein the second rewriting process is implemented based on the rewriting model; A navigation broadcast text is selected from the at least one first broadcast text and the at least one second broadcast text.
2. The method according to claim 1, characterized in that The first rewriting process is performed using the navigation element, the character setting information, and the navigation scene information to obtain at least one first broadcast text, including: Determine, by using the first configuration table, at least one rewriting template corresponding to the character setting information and the navigation scenario information; Obtaining at least one first broadcast text using the navigation element and the at least one rewriting template; The first configuration table includes the correspondence between character setting information, navigation scenario information and rewriting templates.
3. The method according to claim 2, characterized in that The determining, using the first configuration table, at least one rewriting template corresponding to the character setting information and the navigation scenario information comprises: Using the acquired character setting information and the acquired navigation scenario information to query the first configuration table, and using the rewriting template obtained by the query as a candidate rewriting template; selecting the at least one rewriting template from the candidate rewriting templates according to a preset first probability distribution; The first probability distribution includes the selection probability corresponding to each candidate rewriting template.
4. The method according to claim 2, characterized in that: The rewriting template includes a navigation element slot, a fixed speech and a hook word slot; Obtaining at least one first broadcast text by using the navigation element and the at least one rewriting template includes: Using the character setting information to query the second configuration table, obtaining at least one hook word corresponding to the character setting information; Using the navigation element to assign a value to a navigation element slot in the at least one rewriting template, and using the at least one attachment word to assign a value to a attachment word slot in the at least one rewriting template, to obtain at least one first broadcast text; The second configuration table includes the corresponding relationship between the character setting information and the hook words.
5. The method according to claim 4, characterized in that Assigning a value to a hang word slot in the at least one rewriting template using the at least one hang word includes: According to a preset second probability distribution, selecting a hanging word from the at least one hanging word to assign a value to a hanging word slot in the at least one rewriting template; The second probability distribution includes the selection probability corresponding to each attachment word.
6. The method according to claim 5, characterized in that In the second probability distribution, the tone level corresponding to each linking word is negatively correlated with the selection probability.
7. The method according to claim 1, characterized in that The second rewriting process is performed using the navigation element, the character setting information, and the navigation scene information to obtain at least one second broadcast text, including: Determine the announcement words corresponding to the navigation element, and obtain at least one second announcement text corresponding to the announcement words, where the second announcement text is obtained by rewriting the announcement words by a rewriting model using the persona information and the navigation scenario information.
8. The method according to claim 7, characterized in that Acquiring at least one second broadcast text corresponding to the broadcast speech includes: providing the announcement speech, the person setting information and the navigation scene information to a rewriting model, obtaining at least one announcement text obtained by the rewriting model rewriting the announcement speech using the person setting information and the navigation scene information, and obtaining the second announcement text using the at least one announcement text; or The broadcast speech, the character information and the navigation scene information are used to query a rewriting text library to obtain candidate broadcast texts corresponding to the broadcast speech, the character information and the navigation scene information, and at least one second broadcast text is selected from the candidate broadcast texts, wherein the rewriting text library is obtained by rewriting the historical broadcast speech in advance by a rewriting model using the character information and the navigation scene information.
9. The method according to any one of claims 1 to 8, characterized in that Selecting a navigation broadcast text from the at least one first broadcast text and the at least one second broadcast text comprises: According to a preset third probability distribution, selecting one of the at least one first broadcast text and the at least one second broadcast text as the navigation broadcast text; The third probability distribution includes selection probabilities corresponding to the first broadcast text and the second broadcast text respectively.
10. The method according to any one of claims 1 to 8, characterized in that Before executing the method for generating navigation broadcast text, the method further includes: Determining whether the method for generating a navigation broadcast text is used for the prompt point according to a preset fourth probability distribution, and if yes, continuing to execute the method for generating a navigation broadcast text; The fourth probability distribution includes the probability of each prompt point adopting the method for generating the navigation broadcast text.
11. The method according to any one of claims 1 to 8, characterized in that The navigation element is determined according to the navigation route; The navigation scenario is determined based on at least one of a user's location and an environment; The character setting information is determined based on the role information selected by the user.
12. A navigation broadcast method, characterized in that: The method comprises: Obtaining a navigation broadcast voice corresponding to a navigation broadcast text, wherein the navigation broadcast text is generated by the server using the method according to any one of claims 1 to 11; Play the navigation announcement voice.
13. A device for generating navigation broadcast text, characterized in that: The device comprises: An information acquisition unit, configured to acquire navigation elements, character information, and navigation scene information corresponding to the prompt point; A first rewriting unit is configured to perform a first rewriting process using the navigation element, the character setting information and the navigation scene information to obtain at least one first broadcast text, wherein the first rewriting process is implemented based on a rewriting template; A second rewriting unit is configured to perform a second rewriting process using the navigation element, the character setting information and the navigation scene information to obtain at least one second broadcast text, wherein the second rewriting process is implemented based on the rewriting model; The text selection unit is configured to select a navigation broadcast text from the at least one first broadcast text and the at least one second broadcast text.
14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.
15. An electronic device, characterized in that: include: one or more processors; as well as A memory associated with the one or more processors, the memory being used to store program instructions, wherein the program instructions, when read and executed by the one or more processors, execute the steps of the method described in any one of claims 1 to 12.
16. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.
