Playing method and device of vehicle-mounted broadcaster

Abstract

The invention discloses a playing method and device of a vehicle-mounted player. Comprising the following steps: in response to an audio acquisition instruction for a vehicle-mounted player, acquiring current display content of a preset application program of target vehicle-mounted equipment; performing theme recognition and keyword extraction on the current display content to obtain a target theme and a target keyword; based on the target theme and the target keyword, obtaining a recommended text associated with the current display content; and generating a corresponding target broadcaster script based on the recommendation text, generating a target audio of the vehicle-mounted broadcaster based on the target broadcaster script, and playing the target audio through the target vehicle-mounted device. Through the technical scheme provided by the invention, the content watched by the user can be analyzed, the detailed information required to be listened by the user can be accurately obtained, and the finally generated broadcasting audio is ensured to highly focus on the user demand.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of artificial intelligence, and particularly relates to a vehicle-mounted podcast playing method and device. BACKGROUND

[0002] With the continuous development of vehicle intelligent applications, vehicle-mounted podcasts have become an important way for people to obtain information and entertainment in commuting, long-distance driving and other scenarios. Users can listen to news information, audio books and other content through intelligent cockpit devices.

[0003] At present, vehicle-mounted podcasts lack intelligent playing methods, and users need to frequently perform touch operations based on a search function, which not only makes it difficult to quickly find podcast content of interest to the user, but also distracts the attention of the driver during driving, causing safety hazards. SUMMARY

[0004] Embodiments of the application provide a vehicle-mounted podcast playing method and device, which can analyze user viewing content, accurately obtain detailed information required by the user to listen to, and ensure that the finally generated podcast audio is highly focused on user needs.

[0005] Other characteristics and advantages of the application will become apparent from the following detailed description, or will be learned by practice of the application.

[0006] According to a first aspect of an embodiment of the application, a vehicle-mounted podcast playing method is provided, comprising: in response to an audio acquisition instruction of a vehicle-mounted podcast, acquiring current display content of a preset application program of a target vehicle-mounted device; performing theme recognition and keyword extraction on the current display content to obtain a target theme and a target keyword; based on the target theme and the target keyword, acquiring recommended text associated with the current display content; generating a corresponding target podcast script based on the recommended text, generating a target audio of the vehicle-mounted podcast based on the target podcast script, and playing the target audio through the target vehicle-mounted device.

[0007] In some embodiments of the application, based on the foregoing scheme, in response to the audio acquisition instruction of the vehicle-mounted podcast, the method comprises: displaying the preset application program and a podcast control on the display interface of the target vehicle-mounted device; wherein the podcast control is located on the upper layer of the preset application program; in response to a trigger operation of the podcast control, generating the audio acquisition instruction of the vehicle-mounted podcast.

[0008] In some embodiments of the application, based on the foregoing scheme, a play control sub-control is displayed on the podcast control, and the method further comprises: In the process of playing the target audio, the display state of the play control sub-control is switched to a playing state; In response to a triggering operation on the play control sub-control, the playing of the target audio is paused, and the display state of the play control sub-control is switched to a pause state.

[0009] In some embodiments of the present application, based on the foregoing scheme, a progress adjustment sub-control is displayed on the podcast control, and the progress adjustment sub-control displays play progress information. The method further comprises: In response to a progress adjustment operation on the progress adjustment sub-control, the play progress information is updated, and the target audio is played based on the updated play progress information.

[0010] In some embodiments of the present application, based on the foregoing scheme, the current display content of the preset application of the target vehicle-mounted device is obtained, comprising: If the display interface of the target vehicle-mounted device displays web page content of the preset application, the text data to which the web page content belongs is obtained based on the uniform resource locator corresponding to the web page content, and the text data is taken as the current display content; If the display interface of the target vehicle-mounted device displays non-web page content of the preset application, the current display content is obtained by performing text recognition on the non-web page content.

[0011] In some embodiments of the present application, based on the foregoing scheme, the current display content is subjected to theme recognition and keyword extraction to obtain a target theme and a target keyword, comprising: Based on the text type of the current display content, a first prompt information is obtained; The first prompt information and the current display content are input into a large language model to obtain the target theme and the target keyword output by the large language model.

[0012] In some embodiments of the present application, based on the foregoing scheme, based on the target theme and the target keyword, a recommended text associated with the current display content is obtained, comprising: Based on the target theme and the target keyword, at least two candidate texts are searched and obtained; Based on the text similarity between the current display content and the at least two candidate texts, a recommended text associated with the current display content is determined from the at least two candidate texts.

[0013] In some embodiments of the present application, based on the foregoing scheme, based on the text similarity between the current display content and the at least two candidate texts, a recommended text associated with the current display content is determined from the at least two candidate texts, comprising: The current display content and the at least two candidate texts are respectively subjected to feature extraction to obtain a first text vector of the current display content and a second text vector of the at least two candidate texts. For each candidate text, based on the first text vector and the second text vector of the candidate text, determine the text similarity corresponding to the candidate text; Based on the text similarity corresponding to each candidate text, determine the recommended text associated with the current display content from at least two candidate texts.

[0014] In some embodiments of the present application, based on the foregoing scheme, the corresponding target podcast script is generated based on the recommended text, comprising: Input the second prompt information for the podcast script and the recommended text into the large language model to obtain the target podcast script output by the large language model; wherein the target podcast script includes podcast role information, and the podcast role information includes the number of podcast roles and the basic parameters of each podcast role.

[0015] In some embodiments of the present application, based on the foregoing scheme, the target audio of the in-vehicle podcast is generated based on the target podcast script, comprising: Based on the target podcast script, generate the initial audio of each podcast role respectively; Synthesize the initial audio of each podcast role to obtain the target audio of the in-vehicle podcast.

[0016] In some embodiments of the present application, based on the foregoing scheme, for each podcast role, the initial audio of the podcast role is generated based on the target podcast script, comprising: In response to the parameter adjustment instruction of the podcast role, update the basic parameters of the podcast role; wherein the basic parameters include at least one of tone parameter, speech rate parameter and emotion parameter; Generate the initial audio of the podcast role based on the target podcast script and the updated basic parameters of the podcast role through the voice synthesis model.

[0017] According to a second aspect of the embodiments of the present application, a playing device of an in-vehicle podcast is provided, comprising: The content acquisition module is configured to acquire the current display content of the preset application program of the target vehicle-mounted device in response to the audio acquisition instruction of the in-vehicle podcast; The information extraction module is configured to perform theme recognition and keyword extraction on the current display content to obtain a target theme and a target keyword; The text determination module is configured to acquire the recommended text associated with the current display content based on the target theme and the target keyword; The audio playing module is configured to generate a corresponding target podcast script based on the recommended text, generate a target audio of the in-vehicle podcast based on the target podcast script, and play the target audio through the target vehicle-mounted device.

[0018] According to a third aspect of the embodiments of the present application, a computer readable storage medium is provided, and the computer readable storage medium stores computer program instructions. When the computer program instructions are loaded and executed by a processor, the steps of the method according to any one of the first aspect are implemented.

[0019] According to a fourth aspect of the embodiments of the present application, an electronic device is provided, which comprises a memory and a processor. The memory stores computer programs, and when the processor executes the computer programs, the steps of the method according to any one of the first aspect are implemented.

[0020] According to a fifth aspect of the embodiments of the present application, a vehicle is provided, which comprises a target vehicle-mounted device. When the function of the vehicle-mounted podcast is implemented on the target vehicle-mounted device, the steps of the method according to any one of the first aspect are implemented.

[0021] According to a sixth aspect of the embodiments of the present application, a computer program product is provided, which comprises a computer program. When the computer program is executed by a processor, the steps of the method according to any one of the first aspect are implemented.

[0022] In the present application, in response to an audio acquisition instruction of a vehicle-mounted podcast, the current display content of a preset application of a target vehicle-mounted device is acquired. The current display content is subjected to subject identification and keyword extraction to obtain a target subject and a target keyword. Based on the target subject and the target keyword, a recommended text associated with the current display content is acquired. A target podcast script corresponding to the recommended text is generated based on the recommended text, a target audio of the vehicle-mounted podcast is generated based on the target podcast script, and the target audio is played through the target vehicle-mounted device. Through the technical solution provided in the present application, the user's viewing content can be analyzed, and detailed information required by the user to listen to can be accurately acquired, so that the finally generated podcast audio is highly focused on the user's demand.

[0023] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0024] The drawings herein are incorporated into the specification and form part of the specification, show embodiments consistent with the present application, and together with the specification serve to explain the principles of the present application. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. In the drawings: Figure 1 A scene schematic diagram of a vehicle-mounted podcast playing method to which the embodiments of the present application can be applied is shown; Figure 2 A flowchart of a vehicle-mounted podcast playing method in the embodiments of the present application is shown; Figure 3 A flowchart is shown in the embodiment of the present application, which is responsive to an audio acquisition instruction; Figure 4 A flowchart is shown in the embodiment of the present application, which is for playing control of target audio; Figure 5 A basic style of a podcast control in the embodiment of the present application is shown; Figure 6 A flowchart is shown in the embodiment of the present application, which is for determining a target theme and a target keyword; Figure 7 A flowchart is shown in the embodiment of the present application, which is for determining a recommended text; Figure 8 A flowchart is shown in the embodiment of the present application, which is for generating target audio; Figure 9 Another flowchart is shown in the embodiment of the present application, which is for a playing method of an in-vehicle podcast; Figure 10 A block diagram is shown in the embodiment of the present application, which is for a playing device of an in-vehicle podcast; Figure 11 A structural schematic diagram of an electronic device in the embodiment of the present application is shown. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0026] In addition, the described features, structures or characteristics can be combined in any suitable way in one or more embodiments. In the following description, many specific details are provided to give a full understanding of the embodiments of the present application. However, those skilled in the art will realize that the technical solutions of the present application can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. can be used. In other cases, well-known methods, devices, implementations or operations are not shown or described in detail to avoid obscuring the aspects of the present application.

[0027] The block diagrams shown in the drawings are only functional entities, which do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0028] The flowchart shown in the drawings is only an exemplary illustration, and does not necessarily include all contents and operations / steps, nor does it necessarily have to be executed in the order described. For example, some operations / steps can be further decomposed, and some operations / steps can be combined or partially combined, so the actual execution order can be changed according to the actual situation.

[0029] In order for those skilled in the art to better understand the present application, first, in combination with Figure 1 The application scenario involved in the present application is briefly described.

[0030] Referring to Figure 1 , a scene schematic diagram of a playing method of an in-vehicle podcast to which embodiments of the present application can be applied is shown.

[0031] The vehicle-end controller 102 can obtain the current display content of the preset application program of the target in-vehicle device in response to the audio acquisition instruction of the in-vehicle podcast, and then perform theme recognition and keyword extraction on the current display content to obtain a target theme and a target keyword. Based on the target theme and the target keyword, the server 104 obtains a recommended text associated with the current display content, and further generates a target podcast script based on the recommended text, generates a target audio of the in-vehicle podcast based on the target podcast script, and controls the target in-vehicle device to play the target audio.

[0032] Among them, the server 104 can be implemented by an independent server or a server cluster composed of multiple servers, and the vehicle-end controller 102 can search for multiple candidate texts by accessing the server, and then select the required recommended text from the multiple candidate texts. The server 104 can also integrate a data storage system for storing different types of candidate texts, and the data storage system can also be placed on the cloud or other network servers.

[0033] In an exemplary embodiment, referring to Figure 2 , a flowchart of a playing method of an in-vehicle podcast in embodiments of the present application is shown, which is described in detail as follows: Step 201, in response to an audio acquisition instruction of an in-vehicle podcast, obtaining the current display content of a preset application program of a target in-vehicle device.

[0034] Among them, the target in-vehicle device refers to a smart cockpit device with display function on the vehicle, such as a central control display screen, a copilot entertainment screen, etc. Multiple application programs can be run on the target in-vehicle device, and the application program currently displayed on the display interface of the target in-vehicle device is the preset application program.

[0035] Optionally, not all applications can be preset applications, for example, when the currently displayed application on the display interface of the target vehicle-mounted device is a functional application such as vehicle settings, phone, etc., it is generally not a preset application. For all applications running on the target vehicle-mounted device, the preset application is feasible.

[0036] The vehicle-side controller can obtain the current display content of the preset application of the target vehicle-mounted device in response to the audio acquisition instruction of the vehicle-mounted podcast. For example, the user clicks the podcast control on the target vehicle-mounted device to trigger the audio acquisition instruction; the user operates the associated terminal of the target vehicle-mounted device to trigger the audio acquisition instruction; the user triggers the audio acquisition instruction through voice, such as voice information such as "play podcast", and the target vehicle-mounted device can recognize the voice information of the user, and then generate the audio acquisition instruction.

[0037] Step 202, subject recognition and keyword extraction are performed on the current display content to obtain a target subject and a target keyword.

[0038] A lightweight deep learning model is deployed on the vehicle-side controller, which can perform subject recognition and keyword extraction on the current display content to obtain a target subject and a target keyword corresponding to the current display content. The target subject is generally a sentence used to summarize the basic information of the current display content, and the target keyword is a key information extracted from the current display content, which is generally one or more words.

[0039] Optionally, the vehicle-side controller can also transmit the current display content to the server to perform subject recognition and keyword extraction on the current display content through the deep learning model deployed on the server to obtain more rich and accurate target subjects and target keywords.

[0040] Step 203, based on the target subject and the target keyword, a recommended text associated with the current display content is obtained.

[0041] Based on search engine technology, texts related to the target subject and the target keyword are automatically searched, such as news information, encyclopedic knowledge, etc., and then the recommended text is determined from the related texts, wherein the recommended text can be a text with the highest degree of association with the current display content, or a text most suitable for generating the target audio of the vehicle-mounted podcast.

[0042] Optionally, the target subject and the target keyword are searched simultaneously to obtain related texts; the target subject is searched first, and then the target keyword is further searched to obtain related texts.

[0043] Optionally, the target subject and the target keyword are searched in a dynamically updated text database to obtain related texts.

[0044] At step 204, the target podcast script corresponding to the recommended text is generated based on the recommended text, the target audio of the in-vehicle podcast is generated based on the target podcast script, and the target audio is played through the target in-vehicle device.

[0045] The podcast script needs to meet certain format specifications, and the recommended text is converted into the target podcast script according to the format specifications. Then, the target audio corresponding to the target podcast script is generated and automatically played on the target in-vehicle device.

[0046] For example, the recommended text is analyzed by a deep learning model to determine the summary abstract of the recommended text, and then all the summary abstracts are organized, generated, and arranged to obtain the target podcast script.

[0047] For example, a speech synthesis model is also deployed on the vehicle-side controller, which can generate the target audio corresponding to the target podcast script. The speech synthesis model can be a TTS (Text To Speech) model.

[0048] In this application, in response to an audio acquisition instruction of the in-vehicle podcast, the current display content of a preset application program of a target in-vehicle device is acquired; the target theme and target keywords are obtained by subject recognition and keyword extraction on the current display content; the recommended text associated with the current display content is acquired based on the target theme and the target keywords; the target podcast script corresponding to the recommended text is generated based on the recommended text, the target audio of the in-vehicle podcast is generated based on the target podcast script, and the target audio is played through the target in-vehicle device. Through the technical solution provided in this application, the user's viewing content can be analyzed, and detailed information required by the user to listen to can be accurately acquired, ensuring that the finally generated podcast audio is highly focused on the user's needs.

[0049] Based on the above embodiments, in one exemplary embodiment, referring to Figure 3 , a visual method for responding to an audio acquisition instruction in an embodiment of the present application is shown, which specifically includes: At step 301, a preset application program and a podcast control are displayed on a display interface of a target in-vehicle device.

[0050] The podcast control is located above the preset application program, and the podcast control can be a floating interactive control. The position of the podcast control on the display interface can be fixed or draggable. When the user opens the preset application program on the display interface of the target in-vehicle device, the podcast control automatically appears above the preset application program.

[0051] Optionally, when the user does not use the in-vehicle podcast function for a long time, the podcast control can be temporarily hidden, such as completely hidden or automatically hidden at the boundary. When it is needed to use, the podcast control is displayed on the display interface again.

[0052] At step 302, in response to the triggering operation on the podcast control, an audio acquisition instruction of the in-vehicle podcast is generated.

[0053] For example, the user clicks the podcast control to complete the triggering operation on the podcast control, so that the vehicle-side controller generates the audio acquisition instruction of the in-vehicle podcast in response to the triggering operation.

[0054] In this application, a display mode of the podcast control is provided, which can trigger the step of playing the in-vehicle podcast by one key through simple interaction. The automatic process reduces the screen interaction steps of the user, and the user does not need to manually input a search keyword or jump to an application. The target audio of the in-vehicle podcast is seamlessly acquired while using the preset application, which greatly simplifies the information acquisition process, helps to reduce the operation burden of the user, improves the driving safety, and improves the information acquisition efficiency and user experience during driving.

[0055] On the basis of the above-mentioned embodiments, in an exemplary embodiment, a play control sub-control is displayed on the podcast control, as shown in Figure 4 , a visual method for playing control of the target audio in the embodiment of the application is shown, which specifically includes: At step 401, in the process of playing the target audio, the display state of the play control sub-control is switched to the playing state.

[0056] The play control sub-control on the podcast control can be displayed as "play / pause", and the display state includes the playing state and the pause state. When the target audio is automatically played by the target in-vehicle device, the display state of the play control sub-control is automatically switched to the playing state.

[0057] At step 402, in response to the triggering operation on the play control sub-control, the target audio is paused, and the display state of the play control sub-control is switched to the pause state.

[0058] The user clicks the play control sub-control to complete the triggering operation on the play control sub-control, and the target audio of the in-vehicle podcast currently being played is paused. The display state of the play control sub-control is automatically switched to the pause state.

[0059] Optionally, when the display state of the play control sub-control is the pause state, the user can click the play control sub-control to continue to start playing the target audio of the in-vehicle podcast. The display state of the play control sub-control is automatically switched to the playing state.

[0060] In this application, the user can control the playing of the target audio through simple interaction, which is helpful to improve the user experience.

[0061] On the basis of the above-mentioned embodiments, in one exemplary embodiment, a progress adjustment sub-control is displayed on the podcast control, and the progress adjustment sub-control displays the play progress information. The visual method for adjusting the progress of the target audio in the embodiments of the application specifically includes: In response to the progress adjustment operation on the progress adjustment sub-control, the play progress information is updated, and the target audio is played based on the updated play progress information.

[0062] The progress adjustment sub-control on the podcast control can be displayed as a progress bar, and the progress bar has the play progress information. For example, when the target audio of the in-vehicle podcast is not acquired, the progress bar is displayed as “--”. When the target audio of the in-vehicle podcast is started to be played, the progress bar displays the play time of the target audio, in detail to each second.

[0063] The user clicks the progress adjustment sub-control, updates the play progress information of the target audio to the play time corresponding to the clicked position, and continues to start playing the target audio of the in-vehicle podcast from the play time corresponding to the clicked position. The user drags the progress adjustment sub-control, updates the play progress information of the target audio to the play time corresponding to the dragged position, and continues to start playing the target audio of the in-vehicle podcast from the play time corresponding to the dragged position.

[0064] In the present application, the user can simply interact to adjust the progress of the target audio, which is beneficial to improve the user experience.

[0065] Referring to Figure 5 , a basic style of a podcast control is shown, and the display interface of the target in-vehicle device displays a preset application and a podcast control. The podcast control can be a long strip-shaped control, and the left side is a play control sub-control, and the right side is a progress adjustment sub-control.

[0066] Optionally, when the target audio of the in-vehicle podcast is not acquired, the podcast control is a circular control. When the target audio of the in-vehicle podcast is started to be played, the podcast control is a long strip-shaped control, including the play control sub-control and the progress adjustment sub-control.

[0067] On the basis of the above-mentioned embodiments, in one exemplary embodiment, the way of acquiring the current display content specifically includes: If the display interface of the target in-vehicle device displays the web page content of the preset application, the text data to which the web page content belongs is acquired based on the uniform resource locator corresponding to the web page content, and the text data is taken as the current display content. If the display interface of the target in-vehicle device displays the non-web page content of the preset application, the current display content is obtained by performing text recognition on the non-web page content.

[0068] The webpage content is an in-application webpage, such as a browser or a preset application embedded webpage. When the webpage content is displayed on the display interface, a uniform resource locator (URL) link corresponding to the webpage content can be obtained, and all text data to which the webpage content belongs can be obtained based on a crawler technology. In other words, the webpage content displayed on the display interface of the target vehicle-mounted device is only part of the text data, and all text data to which the webpage content belongs can be obtained.

[0069] When the non-webpage content is displayed on the display interface, all text data can be obtained based on screen recognition and an optical character recognition (OCR) technology.

[0070] In this application, the webpage content and the non-webpage content are distinguished, all text data to which the webpage content belongs is obtained, and all text data of the non-webpage content is obtained through text recognition. More detailed and accurate current display content can be obtained, thereby improving the accuracy of determining the content of interest of the user based on the current display content.

[0071] On the basis of the above-mentioned embodiments, in an exemplary embodiment, referring to Figure 6 , a manner of determining a target topic and a target keyword in the embodiments of the application is shown, and specifically includes: In step 601, first prompt information corresponding to the text type of the current display content is obtained.

[0072] The first prompt information is a prompt word of a large language model, and different first prompt information can be used for different text types of the current display content.

[0073] For example, when the text type of the current display content is news information, a preset first prompt information for analyzing news information is obtained, which is represented as "please analyze the time, place, and subject of the news information, summarize the core event in one sentence, and list the keywords reflecting the core content"; when the text type of the current display content is encyclopedic knowledge, a preset first prompt information for analyzing encyclopedic knowledge is obtained, which is represented as "please analyze the concept and characteristics of the encyclopedic knowledge, summarize the concept definition in one sentence, and list the keywords reflecting the concept characteristics".

[0074] In step 602, the first prompt information and the current display content are input into the large language model to obtain a target topic and a target keyword output by the large language model.

[0075] The first prompt information and the current display content are input into the trained large language model, analyzed by the large language model, and a target topic and a target keyword of the current display content are obtained.

[0076] For example, the text type of the current display content is news information, the target theme obtained is "Shenzhou 18 astronaut crew returned safely on November 4, and the mission was a complete success", and the target keywords obtained are "Shenzhou 18", "safe return", "extravehicular activity", and "mission success"; the text type of the current display content is encyclopedic knowledge, and the target theme obtained is "Artificial intelligence is a branch of computer science, aiming to study, develop, and apply systems for simulating, extending, and expanding human intelligence", and the target keywords obtained are "machine learning", "computer vision", and "big data driven".

[0077] In the present application, a way of determining a target theme and target keywords of current display content by a large language model is provided, which can improve the efficiency and accuracy of the determined target theme and target keywords. On this basis, different first prompt information is used for current display content of different text types, which can further fit the text features of the current display content and improve the quality of the text analysis result.

[0078] On the basis of the above-mentioned embodiments, in an exemplary embodiment, referring to Figure 7 , a way of determining recommended text in the embodiments of the present application is shown, which specifically includes: Step 701, searching at least two candidate texts based on the target theme and the target keywords.

[0079] Based on search engine technology, texts related to the target theme and the target keywords are automatically searched, and the search results are read by URL link to obtain a plurality of candidate texts.

[0080] Optionally, the number of candidate texts is preset, such as 5. Texts related to the target theme and the target keywords are searched, and the first 5 related texts obtained by the search are taken as candidate texts.

[0081] Step 702, determining a recommended text associated with the current display content from the at least two candidate texts based on the text similarity between the current display content and the at least two candidate texts.

[0082] The text similarity between the current display content and each candidate text is calculated, and based on the text similarity, the candidate text with the highest degree of association with the current display content is selected from all candidate texts as the recommended text, i.e., the candidate text with the highest text similarity is taken as the recommended text.

[0083] Optionally, cosine similarity matching and word vector similarity are used to calculate the text similarity between the current display content and each candidate text.

[0084] Optionally, feature extraction is performed on the current display content and the at least two candidate texts respectively to obtain a first text vector of the current display content and a second text vector of the at least two candidate texts; for each candidate text, a text similarity corresponding to the candidate text is determined based on the first text vector and the second text vector of the candidate text; and based on the text similarity corresponding to each candidate text, a recommended text associated with the current display content is determined from the at least two candidate texts.

[0085] The first text vector is used to represent the text features of the current display content, and the second text vector of each candidate text is used to represent the text features of the candidate text.

[0086] In the present application, a way of determining a recommended text associated with the current display content by calculating a text similarity is provided, which can find a candidate text with the highest degree of association with the current display content from multiple candidate texts, perceive the user's intention, and find the content that the user is most interested in at present.

[0087] Based on the above-mentioned embodiments, in an exemplary embodiment, the way of generating the target podcast script specifically includes: inputting the second prompt information for the podcast script and the recommended text into the large language model to obtain a target podcast script output by the large language model; wherein the target podcast script includes podcast role information, and the podcast role information includes the number of podcast roles and basic parameters of each podcast role.

[0088] The second prompt information is a prompt word of the large language model, and is preset for generating the podcast script. Exemplarily, the second prompt information is represented as "please generate a podcast script based on the recommended text, and multiple podcast roles can appear in the podcast script, and each podcast role has a different style".

[0089] The second prompt information and the recommended text are input into the trained large language model together, and the large language model is analyzed to obtain a target podcast script corresponding to the recommended text. The podcast role refers to a virtual role that presents content through voice, and the basic parameters of different podcast roles are different, and the target podcast script describes which voice of which podcast role presents each paragraph of speech.

[0090] In the present application, a way of generating a target podcast script through a large language model is provided, which can quickly generate a personalized target podcast script and ensure the playing effect of the in-vehicle podcast.

[0091] Based on the above-mentioned embodiments, in an exemplary embodiment, referring to Figure 8 , a way of generating a target audio in the embodiments of the present application is shown, which specifically includes: Step 801, based on the target podcast script, generating initial audio of each podcast character respectively.

[0092] The initial audio of each podcast character is generated through a voice synthesis model. For example, the complete target podcast script is split into independent text segments according to the podcast characters, and then independent initial audio is generated for each podcast character.

[0093] Optionally, in response to a parameter adjustment instruction for a podcast character, the basic parameters of the podcast character are updated; wherein the basic parameters include at least one of tone parameter, speech rate parameter and emotion parameter; through a voice synthesis model, based on the target podcast script and the updated basic parameters of the podcast character, the initial audio of the podcast character is generated.

[0094] The user can adjust the basic parameters of the podcast character at any time based on his own preferences, wherein the updateable basic parameters include but are not limited to tone parameter, speech rate parameter and emotion parameter, tone parameter such as young and energetic male voice, mature and stable female voice, neutral and magnetic voice, speech rate parameter such as medium-fast, smooth, slow and appropriate pause, and emotion parameter such as excitement, seriousness and mystery.

[0095] Step 802, synthesizing the initial audio of each podcast character to obtain the target audio of the in-vehicle podcast.

[0096] The initial audio of all podcast characters is aligned and edited in multiple tracks, and appropriate effect processing is performed. For example, according to the timeline in the target podcast script, the initial audio of each podcast character is accurately aligned, the volume and frequency of each initial audio are balanced, and background music is embedded at appropriate positions to obtain the target audio of the in-vehicle podcast that is more comfortable for the user to listen to.

[0097] In this application, the target podcast script is converted into target audio, realizing the combination of visual browsing and auditory listening, ensuring the quality of the generated target audio of the in-vehicle podcast, and meeting the user's demand for multi-modal information acquisition in the driving scenario.

[0098] On the basis of the above embodiments, in an exemplary embodiment, referring to Figure 9 , another flowchart of the playing method of the in-vehicle podcast in the embodiment of the application is shown, which is described in detail as follows: Step 901, in response to an audio acquisition instruction for an in-vehicle podcast, determining whether the content displayed on the display interface of the target in-vehicle device belongs to web page content.

[0099] If yes, step 902 is executed; if no, step 903 is executed.

[0100] Step 902: Based on the Uniform Resource Locator (URL) corresponding to the webpage content, obtain the text data to which the webpage content belongs, and use the text data as the currently displayed content.

[0101] Step 903: Perform text recognition on the non-web page content to obtain the currently displayed content.

[0102] Step 904: Based on the text type of the currently displayed content, obtain the corresponding first prompt information, input the first prompt information and the currently displayed content into the large language model, and obtain the target topic and target keywords output by the large language model.

[0103] Step 905: Based on the target topic and target keywords, at least two candidate texts are obtained. Features are extracted from the currently displayed content and the at least two candidate texts to obtain the first text vector of the currently displayed content and the second text vectors of the at least two candidate texts. For each candidate text, the text similarity corresponding to the candidate text is determined based on the first text vector and the second text vector of the candidate text. Based on the text similarity corresponding to each candidate text, the recommended text associated with the currently displayed content is determined from the at least two candidate texts.

[0104] Step 906: Input the second cue information and recommendation text for the podcast script into the large language model to obtain the target podcast script output by the large language model.

[0105] Step 907: Based on the target podcast script, generate the initial audio for each podcast character, and synthesize the initial audio for each podcast character to obtain the target audio for the in-vehicle podcast.

[0106] Step 908: Play the target audio through the target vehicle-mounted device.

[0107] This application is able to analyze the content viewed by users, accurately obtain the detailed information that users need to listen to, and ensure that the final podcast audio is highly focused on user needs.

[0108] The following describes an embodiment of the apparatus described in this application, which can be used to execute the in-vehicle podcast playback method described in the above embodiments of this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the in-vehicle podcast playback method described above.

[0109] See Figure 10 The diagram shows a block diagram of a car podcast playback device 1000 according to an embodiment of this application, specifically including: The content acquisition module 1001 is used to acquire the currently displayed content of the preset application of the target vehicle device in response to the audio acquisition command of the in-vehicle podcast. The information extraction module 1002 is configured to perform subject identification and keyword extraction on the current display content to obtain a target subject and a target keyword. The text determination module 1003 is configured to acquire a recommended text associated with the current display content based on the target subject and the target keyword. The audio playing module 1004 is configured to generate a target podcast script corresponding to the recommended text, generate a target audio of the in-vehicle podcast based on the target podcast script, and play the target audio through the target in-vehicle device.

[0110] In an example embodiment, on the basis of the above-mentioned embodiment, the content acquisition module 1001 is specifically configured to display a preset application and a podcast control on a display interface of the target in-vehicle device; the podcast control is located above the preset application; and in response to a triggering operation on the podcast control, an audio acquisition instruction of the in-vehicle podcast is generated.

[0111] In an example embodiment, on the basis of the above-mentioned embodiment, a play control sub-control is displayed on the podcast control, and the in-vehicle podcast playing apparatus 1000 further includes a play control module, which is specifically configured to switch a display state of the play control sub-control to a playing state in the process of playing the target audio; in response to a triggering operation on the play control sub-control, the playing of the target audio is paused, and the display state of the play control sub-control is switched to a pause state.

[0112] In an example embodiment, on the basis of the above-mentioned embodiment, a progress adjustment sub-control is displayed on the podcast control, and the progress adjustment sub-control displays play progress information, and the in-vehicle podcast playing apparatus 1000 further includes a progress adjustment module, which is specifically configured to update the play progress information in response to a progress adjustment operation on the progress adjustment sub-control, and play the target audio based on the updated play progress information.

[0113] In an example embodiment, on the basis of the above-mentioned embodiment, the content acquisition module 1001 includes: A first acquisition unit is configured to, if the display interface of the target in-vehicle device displays web page content of the preset application, acquire text data to which the web page content belongs based on a uniform resource locator corresponding to the web page content, and take the text data as the current display content. A second acquisition unit is configured to, if the display interface of the target in-vehicle device displays non-web page content of the preset application, perform text recognition on the non-web page content to obtain the current display content.

[0114] In an example embodiment, on the basis of the above-mentioned embodiment, the information extraction module 1002 includes: An information acquisition unit is configured to acquire corresponding first prompt information based on a text type of the current display content. The information extraction unit is configured to input the first prompt information and the current display content into the large language model to obtain a target topic and a target keyword output by the large language model.

[0115] In an example embodiment, the text determination module 1003 described above includes the following on the basis of the above embodiment: The text searching unit is configured to search for at least two candidate texts based on the target topic and the target keyword. The text determination unit is configured to determine a recommended text associated with the current display content from the at least two candidate texts based on a text similarity between the current display content and the at least two candidate texts.

[0116] In an example embodiment, the text determination unit described above includes the following on the basis of the above embodiment: The vector extraction subunit is configured to perform feature extraction on the current display content and the at least two candidate texts respectively to obtain a first text vector of the current display content and second text vectors of the at least two candidate texts. The similarity calculation subunit is configured to determine, for each candidate text, a text similarity corresponding to the candidate text based on the first text vector and the second text vector of the candidate text. The text determination subunit is configured to determine a recommended text associated with the current display content from the at least two candidate texts based on the text similarity corresponding to each candidate text.

[0117] In an example embodiment, the audio playing module 1004 described above includes the following on the basis of the above embodiment: The script generation unit is configured to input the second prompt information for the podcast script and the recommended text into the large language model to obtain a target podcast script output by the large language model; the target podcast script includes podcast role information, and the podcast role information includes a number of podcast roles and basic parameters of each podcast role.

[0118] In an example embodiment, the audio playing module 1004 described above includes the following on the basis of the above embodiment: The first audio generation unit is configured to generate initial audio of each podcast role based on the target podcast script. The second audio generation unit is configured to synthesize the initial audio of each podcast role to obtain target audio of the in-vehicle podcast.

[0119] In an example embodiment, the first audio generation unit described above includes the following on the basis of the above embodiment: The parameter updating subunit is configured to update basic parameters of the podcast character in response to a parameter adjustment instruction for the podcast character, wherein the basic parameters include at least one of a tone parameter, a speech speed parameter, and an emotion parameter. The audio generation subunit is configured to generate initial audio of the podcast character based on the target podcast script and the updated basic parameters of the podcast character by using a speech synthesis model.

[0120] Based on the same inventive concept, the embodiments of the present application provide a computer readable storage medium, which stores computer program instructions, and the computer program instructions are loaded and executed by a processor to implement the steps of the playing method of the vehicle-mounted podcast.

[0121] Based on the same inventive concept, the embodiments of the present application provide an electronic device, which is shown in a structural schematic diagram of the electronic device in the embodiments of the present application, and the electronic device includes one or more memories 1104, one or more processors 1102, and at least one computer program stored in the memory 1104 and executable on the processor 1102, and when the processor 1102 executes the computer program, the steps of the playing method of the vehicle-mounted podcast are implemented. Figure 11

[0122] The bus architecture (represented by the bus 1100) can include any number of interconnected buses and bridges, and the bus 1100 links various circuits including one or more processors represented by the processor 1102 and the memory represented by the memory 1104. The bus 1100 can also link various other circuits such as peripheral devices, voltage stabilizers, and power management circuits, which are well known in the art, and thus, further description thereof will not be given herein. The bus interface 1105 provides an interface between the bus 1100 and the receiver 1101 and the transmitter 1103. The receiver 1101 and the transmitter 1103 can be the same element, i.e., a transceiver, which provides a unit for communicating with various other devices on a transmission medium. The processor 1102 is responsible for managing the bus 1100 and general processing, and the memory 1104 can be used to store data used by the processor 1102 in performing operations.

[0123] ​The functions described herein can be implemented in hardware, software executed by a processor, firmware, or any combination thereof. If implemented in software executed by a processor, the functions can be stored on or transferred over as one or more instructions or code on a computer-readable medium. Other examples and implementations are within the scope and spirit of the disclosure and appended claims. For example, due to the nature of software, functions described above can be implemented using software executed by a processor, hardware, firmware, hardwiring, or combinations of any of these. Features implementing functions can also be physically located at various positions, including being distributed such that portions of functions are implemented at different physical locations. Also, as used herein, including in the claims, "or" as used in a list of items prefaced by "comprising" to indicate a disjunctive list means each single item in the list has been recited individually. Thus, such term should be interpreted as "comprising at least one of the listed items." Further, it is to be understood that such aspects can be implemented by entities that are different from those described, such as a middleware layer.

[0124] Based on the same inventive concept, the embodiment of the present application provides a vehicle, comprising a target vehicle-mounted device, when a function of a vehicle-mounted podcast is implemented on the target vehicle-mounted device, the steps of the playing method of the vehicle-mounted podcast are implemented.

[0125] Based on the same inventive concept, the embodiment of the present application provides a computer program product, comprising a computer program, when the computer program is executed by a processor, the steps of the playing method of the vehicle-mounted podcast are implemented.

[0126] In several embodiments provided by the present application, it should be understood that the disclosed technology can be implemented by other means. Among them, the above-mentioned device embodiments are only schematic, for example, the division of the units can be a logical function division, and actual implementation can have another division mode, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed each other can be through some interface, indirect coupling or communication connection between units or modules, which can be electrical or other forms.

[0127] The units described as separate components can or can not be physically separate, and the components of the control device can or can not be physical units, that is, they can be located in one place, or they can be distributed to multiple units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0128] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a mobile hard disk, a magnetic disk or an optical disk and various computer program instruction storage media.

[0129] The above only describes the embodiments of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the scope of claims of the present application.

Claims

1. A method for playing a podcast in a car, characterized in that, The method includes: In response to an audio acquisition command for an in-vehicle podcast, retrieve the currently displayed content of the default application on the target in-vehicle device; The currently displayed content is subjected to topic identification and keyword extraction to obtain the target topic and target keywords; Based on the target topic and the target keywords, obtain the recommended text associated with the currently displayed content; Based on the recommended text, a corresponding target podcast script is generated. Based on the target podcast script, the target audio for the in-vehicle podcast is generated and played through the target in-vehicle device.

2. The method according to claim 1, characterized in that, The response to the audio acquisition instruction for the in-vehicle podcast includes: A preset application and a podcast control are displayed on the display interface of the target vehicle device; wherein the podcast control is located on top of the preset application; In response to a trigger operation on the podcast control, an audio acquisition instruction for the in-vehicle podcast is generated.

3. The method according to claim 2, characterized in that, The podcast control displays a playback control sub-control, and the method further includes: During the playback of the target audio, the display state of the playback control sub-control is switched to playback state; In response to a trigger operation on the playback control sub-control, the playback of the target audio is paused, and the display state of the playback control sub-control is switched to the paused state.

4. The method according to claim 2, characterized in that, The podcast control displays a progress adjustment sub-control, which displays playback progress information. The method further includes: In response to a progress adjustment operation on the progress adjustment sub-control, the playback progress information is updated, and the target audio is played based on the updated playback progress information.

5. The method according to claim 1, characterized in that, The acquisition of the current display content of the preset application of the target vehicle device includes: If the target vehicle device displays web page content of a preset application on its display interface, then based on the Uniform Resource Locator (URL) corresponding to the web page content, the text data to which the web page content belongs is obtained, and the text data is used as the currently displayed content. If the target vehicle device displays non-web page content of a preset application on its display interface, then text recognition is performed on the non-web page content to obtain the currently displayed content.

6. The method according to claim 1, characterized in that, The step of performing topic identification and keyword extraction on the currently displayed content to obtain the target topic and target keywords includes: Based on the text type of the currently displayed content, obtain the corresponding first prompt information; The first prompt information and the currently displayed content are input into the large language model to obtain the target topic and target keywords output by the large language model.

7. The method according to claim 1, characterized in that, The step of obtaining recommended text associated with the currently displayed content based on the target topic and the target keywords includes: Based on the target topic and the target keywords, at least two candidate texts are obtained through the search; Based on the text similarity between the currently displayed content and the at least two candidate texts, recommended text associated with the currently displayed content is determined from the at least two candidate texts.

8. The method according to claim 7, characterized in that, The step of determining recommended text associated with the currently displayed content from the at least two candidate texts based on the text similarity between the currently displayed content and the at least two candidate texts includes: Feature extraction is performed on the currently displayed content and the at least two candidate texts respectively to obtain a first text vector of the currently displayed content and a second text vector of the at least two candidate texts; For each candidate text, the text similarity corresponding to the candidate text is determined based on the first text vector and the second text vector of the candidate text. Based on the text similarity of each candidate text, the recommended text associated with the currently displayed content is determined from the at least two candidate texts.

9. The method according to claim 1, characterized in that, The process of generating the corresponding target podcast script based on the recommended text includes: The second prompt information for the podcast script and the recommended text are input into the large language model to obtain the target podcast script output by the large language model; wherein, the target podcast script includes podcast role information, which includes the number of podcast roles and the basic parameters of each podcast role.

10. The method according to claim 9, characterized in that, The process of generating the target audio for the in-vehicle podcast based on the target podcast script includes: Based on the target podcast script, generate the initial audio for each podcast character; The initial audio of each podcast character is synthesized to obtain the target audio of the in-vehicle podcast.

11. The method according to claim 10, characterized in that, For each podcast character, based on the target podcast script, generate the initial audio for that podcast character, including: In response to a parameter adjustment instruction for the podcast character, the basic parameters of the podcast character are updated; wherein the basic parameters include at least one of timbre parameters, speech rate parameters, and emotion parameters; Using a speech synthesis model, the initial audio of the podcast character is generated based on the target podcast script and the updated basic parameters of the podcast character.

12. A device for playing podcasts in a car, characterized in that, The device includes: The content acquisition module is used to acquire the currently displayed content of the preset application of the target in-vehicle device in response to the audio acquisition command of the in-vehicle podcast. The information extraction module is used to perform topic identification and keyword extraction on the currently displayed content to obtain the target topic and target keywords; The text determination module is used to obtain recommended text associated with the currently displayed content based on the target topic and the target keywords; An audio playback module is used to generate a corresponding target podcast script based on the recommended text, generate the target audio of the in-vehicle podcast based on the target podcast script, and play the target audio through the target in-vehicle device.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when loaded and executed by a processor, implement the steps of the method as described in any one of claims 1-11.

14. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method as described in any one of claims 1-11.

15. A vehicle, characterized in that, The method includes a target in-vehicle device, and when the in-vehicle podcast function is implemented on the target in-vehicle device, the steps of the method as described in any one of claims 1-11 are implemented.

Images