Interaction method, device and equipment based on guest broadcasting system, medium and product

Abstract

The invention discloses an interaction method, device and equipment based on a guest broadcasting system, a medium and a product. The method comprises the following steps: in a process of interacting with a user, based on a preset automatic voice recognition algorithm, performing recognition processing on interaction voice input by the user to obtain target interaction information; according to the target interaction information, under the condition that it is determined that the user does not replace the interaction topic, matching retrieval is conducted in a preset knowledge base according to the original topic and the target interaction information, and feedback information corresponding to the target interaction information is obtained; according to the technical scheme of the invention, voice recognition, large model text generation and voice synthesis technologies can be combined, the voice recognition efficiency can be improved, the user experience can be improved, the user experience can be improved, the user experience can be improved, and the user experience can be improved by combining the voice recognition technology, the large model text generation technology and the voice synthesis technology. Personalized content is better fed back to the user, the interactivity between the guest broadcasting system and the user is improved, and the user experience is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an interactive method, device, equipment, medium and product based on a podcast system. Background Art

[0002] With the development of Internet technology, podcasts are becoming more and more popular as a way of information dissemination and entertainment. However, existing podcast systems are mostly one-way information transmission, lack of interactivity, and slow content update speed, which cannot meet users' needs for personalization and immediacy.

[0003] Therefore, how to combine speech recognition, large model text generation and speech synthesis technology to better provide users with personalized content, increase the interactivity between the podcast system and users, and improve user experience is an urgent problem to be solved. Summary of the invention

[0004] The present invention provides an interactive method, device, equipment, medium and product based on a podcast system, which combines speech recognition, large model text generation and speech synthesis technology to better provide personalized content feedback to users, increase the interactivity between the podcast system and users, and improve user experience.

[0005] According to one aspect of the present invention, there is provided an interactive method based on a podcast system, comprising:

[0006] In the process of interacting with the user, based on the preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain the target interactive information;

[0007] According to the target interaction information, when it is determined that the user has not changed the interaction topic, a matching search is performed in a preset knowledge base according to the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information;

[0008] The target podcast role involved is determined according to the feedback information, and based on the feedback information and the target podcast role and based on a preset large language model, the target podcast dialogue script corresponding to the feedback information is determined to provide feedback to the user.

[0009] According to another aspect of the present invention, there is provided an interactive device based on a podcast system, comprising:

[0010] The recognition module is used to recognize and process the interactive speech input by the user based on a preset automatic speech recognition algorithm during the interaction with the user to obtain the target interactive information;

[0011] A matching module is used to perform matching retrieval in a preset knowledge base based on the original topic and the target interaction information according to the target interaction information, and obtain feedback information corresponding to the target interaction information when it is determined that the user has not changed the interaction topic;

[0012] The feedback module is used to determine the target podcast role involved according to the feedback information, and determine the target podcast script corresponding to the feedback information based on the preset large language model according to the feedback information and the target podcast role, so as to provide feedback to the user.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the interactive method based on the podcast system described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the interactive method based on the podcast system described in any embodiment of the present invention when executed.

[0018] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the interactive method based on the podcast system of any embodiment of the present invention is implemented.

[0019] The technical solution of the embodiment of the present invention is as follows: in the process of interacting with the user, based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain target interaction information; based on the target interaction information, when it is determined that the user has not changed the interactive topic, a matching search is performed in a preset knowledge base based on the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information; based on the feedback information, the target podcast role involved is determined, and based on the feedback information and the target podcast role, the target podcast script corresponding to the feedback information is determined based on a preset large language model to provide feedback to the user. By combining speech recognition, large model text generation and speech synthesis technology, personalized content can be better fed back to the user, the interactivity between the podcast system and the user can be increased, real-time interaction between the user and the podcast system can be achieved, and the user experience can be improved.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 is a flowchart of an interactive method based on a podcast system provided in Embodiment 1 of the present invention;

[0023] Figure 2 is a schematic diagram of the structure of the intelligent podcast system provided by the second embodiment of the present invention;

[0024] Figure 3 is a structural block diagram of an interactive device based on a podcast system provided by Embodiment 3 of the present invention;

[0025] Figure 4 It is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", "target", "candidate", "alternative", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices. The acquisition, storage, use, processing, etc. of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0028] Embodiment 1

[0029] Figure 1 is a flowchart of an interactive method based on a podcast system provided in Embodiment 1 of the present invention; this embodiment is applicable to the case where an intelligent podcast system responds to user needs in the form of a podcast script to provide feedback to the user, and the method can be executed by an interactive device based on a podcast system, and the interactive device based on the podcast system can be implemented in the form of hardware and / or software, and the interactive device based on the podcast system can be configured in an electronic device, such as an intelligent podcast system with interactive functions, wherein the intelligent podcast system may include an AI model service module, a server, and a client, such as Figure 1 As shown, the interactive method based on the podcast system includes:

[0030] S101. In the process of interacting with the user, based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain target interactive information.

[0031] Among them, the intelligent podcast system can interact with the user through the client to obtain the interactive voice input by the user. The automatic speech recognition algorithm (Automatic Speech Recognition, ASR) can be, for example, a method based on a statistical model or a convolutional neural network (Convolutional Neural Networks, CNN) algorithm. Interactive voice refers to the voice input by the user in the client of the intelligent podcast system. The interactive voice can specifically be the voice of the user changing the topic or asking in-depth questions about the current topic. The target interactive information refers to the information obtained by identifying and extracting features from the interactive voice. The target interactive information can be voice instructions or question information. The question information can specifically be information asking questions about the current hot search situation of the topic.

[0032] Optionally, in the process of interacting with the user, based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain the target interactive information, including: interacting with the user based on a preset communication protocol, and in the process of interacting with the user, obtaining the interactive speech input by the user; based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain the target interactive information. The preset communication protocol may be, for example, a Web Socket long connection.

[0033] Optionally, after obtaining the interactive voice input by the user, the interactive voice can be converted into text information based on a preset automatic speech recognition algorithm, and keywords in the text information can be extracted based on natural language processing (NLP) technology to obtain target interactive information.

[0034] Optionally, in the process of interacting with the user, the intelligent podcast system can monitor whether the user inputs new interactive voice through the client based on a preset communication protocol. If it is detected that the user inputs a new interactive voice, the currently being executed operation can be directly interrupted, and the execution of S101-S103 can be returned to perform recognition and processing on the interactive voice input by the user based on a preset automatic speech recognition algorithm, so as to provide feedback on the interactive voice newly input by the user.

[0035] S102: Based on the target interaction information, when it is determined that the user has not changed the interaction topic, a matching search is performed in a preset knowledge base based on the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information.

[0036] Among them, the interactive topic refers to the topic that the intelligent podcast system interactively displays to the user. The interactive topic can be a news report or an interpretation of a famous book, and the present invention does not limit this. The original topic refers to the interactive topic between the intelligent system and the user during the current interaction process. The preset knowledge base refers to the knowledge base pre-configured in the intelligent podcast system. The preset knowledge base can be a real-time updated knowledge base, and the preset knowledge base can store the latest news information and knowledge texts. Feedback information refers to information that responds to the target interactive information.

[0037] Optionally, determining whether the user has changed the interaction topic includes: determining the original topic currently interacting with the user, and performing similarity matching between the target interaction information and the original topic; if the similarity between the target interaction information and the original topic is greater than a preset similarity threshold, determining that the similarity match is successful, and determining that the user has not changed the interaction topic.

[0038] Optionally, the similarity between the target interaction information and the original topic may be determined based on a preset similarity calculation method, such as a cosine similarity calculation method.

[0039] Optionally, if the similarity between the target interaction information and the original topic is less than or equal to a preset similarity threshold, it is determined that the user has changed the interaction topic, and the latest topic can be obtained by extracting information from the target interaction information. Accordingly, a matching search is performed in a preset knowledge base based on the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information, including: a matching search is performed in a preset knowledge base based on the latest topic and the target interaction information to obtain feedback information corresponding to the target interaction information.

[0040] Optionally, a preliminary screening may be performed in a preset knowledge base according to the latest topic to obtain topic knowledge content associated with the latest topic, and matching retrieval may be performed in the topic knowledge content to obtain feedback information corresponding to the target interaction information.

[0041] Optionally, a matching search is performed in a preset knowledge base based on the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information, including: performing a preliminary screening in the preset knowledge base based on the original topic to obtain target knowledge content associated with the original topic; matching and searching in the target knowledge content based on the target interaction information, and generating feedback information corresponding to the target interaction information based on the matching results.

[0042] The feedback information is news information or other knowledge information corresponding to the target interactive information. The target knowledge content refers to the knowledge associated with the original topic stored in the preset knowledge base.

[0043] Optionally, the directory information of the preset knowledge base can be matched with the original topic for similarity to determine the topic directory information related to the original topic, and based on the topic directory information, preliminary screening can be performed in the preset knowledge base to obtain target knowledge content associated with the original topic.

[0044] Optionally, content information having a similarity with the target interactive information greater than a preset similarity threshold may be matched and retrieved from the target knowledge content as feedback information corresponding to the target interactive information.

[0045] Optionally, the knowledge base may be indexed in advance to create an index database, and then search may be performed according to user needs, thereby achieving rapid search of the preset knowledge base by matching the index and the result set.

[0046] S103. Determine the target podcast role involved according to the feedback information, and determine the target podcast script corresponding to the feedback information based on the preset large language model according to the feedback information and the target podcast role, so as to provide feedback to the user.

[0047] Among them, the target podcast role refers to the podcast role that the target shows to the user. For different podcast roles, the intelligent podcast system can use different language styles and expressions to show feedback information to the user. The preset large language model can, for example, be a deep learning model GPT (Generative Pre-Trained Transformer). The target podcast script refers to a script that contains the target podcast role and its corresponding lines. The number of target podcast roles is at least one. For example, the target podcast role can be at least one of the host, guest, expert, reporter, and interviewee, and the target podcast script can be, for example, "Host: xxxx, guest: xxxx".

[0048] Optionally, the feedback information can be divided into different text contents according to the emotional characteristics corresponding to the feedback information, and the corresponding target podcast roles can be matched to each type of text content according to the role characteristics of each candidate podcast role, that is, the target podcast role can be screened and determined from the candidate podcast roles.

[0049] Optionally, according to the feedback information and the target podcast character, based on a preset large language model, a target podcast script corresponding to the feedback information is determined, including: determining the language style and expression corresponding to the target podcast character; according to the language style, expression and feedback information, using a preset large language model, converting the feedback information into a podcast script format, to obtain the target podcast script corresponding to the feedback information.

[0050] The language style refers to a style that can characterize the emotions of the target podcast character, and the voice style can be, for example, positive or negative. The expression can include speech speed, pitch, and volume, such as fast or slow speech, such as saying "so funny" by laughing, such as saying "very noteworthy" by emphasizing.

[0051] Exemplarily, the script format of a podcast may include the following parts: Introduction, where the podcast host introduces himself and the topic of the podcast, introduces the guests, including their background, achievements and experiences related to the interview topic, and finally briefly introduces the main content and purpose of this podcast; Icebreaker may be about the guest's personal life, hobbies and interests to introduce recent news events. Diving Deep, start discussing the core topic of this podcast. It can be the guest's professional field, their work, their opinions or the projects they are working on. Ask in-depth questions and encourage guests to share their insights and experiences. Discussion, have an in-depth conversation with the guests on the topic. It can include different views on the topic, case studies, experience sharing, etc. Encourage guests to share their stories and how these stories are related to the lives of the listeners. Audience Engagement, live podcast, you can invite listeners to ask questions and let the guests answer. Interview street people. Highlights Recap, review the highlights and key information of this podcast. Emphasize the important opinions or interesting stories shared by the guests. Closing: Thank the guest for participating in the interview. Summarize the main points of this podcast. Remind the audience of the next podcast topic or guest to increase their expectations. End with a thought-provoking remark or a lighthearted joke to leave a deep impression on the audience.

[0052] Optionally, the feedback information can be divided into different text contents according to the emotional characteristics corresponding to the feedback information, and a podcast script can be generated based on the format requirements of the podcast script according to the order of different text contents and the target podcast roles corresponding to each text content, so as to obtain the target podcast script corresponding to the feedback information.

[0053] Optionally, corresponding podcast audio can be generated based on the target podcast script based on speech synthesis technology, and feedback can be provided to users based on the podcast audio.

[0054] Optionally, after obtaining the podcast audio, the corresponding ambient sound can be matched to the podcast audio according to the context or user preference, and the ambient sound can be integrated into the podcast audio to feed back the integrated podcast audio to the user. The ambient sound can be, for example, natural sound (such as strong wind and rain), theater audience sound (applause, cheers, whispers, sighs), festival celebrations (firecrackers, festival music, crowd cheers), and suspense sound effects (gloomy, laughter, whispers).

[0055] Optionally, feedback is provided to the user, including: based on a preset speech synthesis algorithm, generating podcast audio corresponding to the target podcast script according to the speech materials corresponding to the target podcast script and the target podcast character; based on a preset communication protocol, feeding back the podcast audio to the user to complete the interaction process with the user.

[0056] The preset speech synthesis algorithm may be an algorithm based on TTS (Text To Speech) technology, such as a speech synthesis algorithm based on a convolutional neural network or a recurrent neural network. The speech material refers to the speech material pre-configured for different podcast roles in the intelligent podcast system. The preset communication protocol may be a web socket long connection.

[0057] Optionally, based on the script content of the target podcast script, it can be determined whether there is a target voice that displays the script content in the voice material corresponding to the target podcast character. If so, the podcast audio corresponding to the target podcast script can be generated based on the target voice. If not, the podcast audio corresponding to the target podcast script can be generated based on the voice material in combination with AI speech synthesis technology.

[0058] For example, a user can ask a new question while playing a podcast audio. If the intelligent podcast system determines that the user has not changed the topic, it can use the large language model to determine a new target podcast script based on the user's new question and the current playback content. For example, host: Today, the xx Olympic Games closed, and team A won a total of 40 gold medals. User: How many gold medals are there in table tennis? (The podcast content and the user's question are processed and retrieved together); the new target podcast script can be: host: Team A won 5 gold medals in the xx Olympic Games, including the men's team.....

[0059] The technical solution of the embodiment of the present invention is as follows: in the process of interacting with the user, based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain target interaction information; based on the target interaction information, when it is determined that the user has not changed the interactive topic, a matching search is performed in a preset knowledge base based on the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information; based on the feedback information, the target podcast role involved is determined, and based on the feedback information and the target podcast role, the target podcast script corresponding to the feedback information is determined based on a preset large language model to provide feedback to the user. By combining speech recognition, large model text generation and speech synthesis technology, personalized content can be better fed back to the user, the interactivity between the podcast system and the user can be increased, real-time interaction between the user and the podcast system can be achieved, and the user experience can be improved.

[0060] Embodiment 2

[0061] Figure 2Schematic diagram of the structure of the intelligent podcast system provided in the second embodiment of the present invention; based on the above embodiment, this embodiment provides a preferred solution for interacting with users through the intelligent podcast system.

[0062] Specifically, Figure 2 As shown, the intelligent podcast system may include an AI model service module, a server, and a client, wherein the AI ​​model service module may include a speech recognition Asr, a large model GPT, and a speech synthesis TTS unit. The speech recognition Asr is used to convert the user's voice input into text information to obtain target interaction information.

[0063] Optionally, the server can obtain the interactive voice sent by the user through the client through the web_socket long connection service (specifically, it can be through command input channel hot search or voice input), and perform voice recognition through the voice recognition Asr of the AI ​​model service module when voice input is detected; the server can also determine the topic content (i.e., target interactive information) based on the results of channel hot search and voice recognition, and feedback to the content retrieval module through the web_socket long connection service. Before performing content retrieval based on the topic content, the content retrieval module can first determine whether the user has new voice input. If there is new input, the retrieval operation is terminated and the operation of obtaining the user's interactive voice is returned. If there is no new input, content retrieval is performed to obtain feedback information, and the large model gpt of the AI ​​model service module is called to organize the feedback information into the target podcast dialogue script (i.e., the script in the "role: dialogue" format).

[0064] It should be noted that Web Socket is used mainly for the following reasons: Real-time: The Web Socket protocol uses a single TCP connection to implement full-duplex communication, so it provides real-time communication, making communication between the server and the client faster, which is very important for radio podcasts because listeners usually want to listen to the program in real time rather than delayed listening. Two-way communication: Web Socket supports two-way communication between the server and the client, which allows listeners to interact with the host in real time during the live broadcast of the podcast. For example, while the radio is playing, the user's speech can be submitted directly, and the virtual host will immediately update the original content to the user's new question, realizing an interactive experience like being on the scene.

[0065] Optionally, public podcast content can be used as a training set to train a large language model. In this way, the richness and stability of the script converted into podcast form can be improved, and a script with lines that conform to the podcast style can be generated based on the user's input text, that is, the target podcast script.

[0066] For example, the AI ​​model service can be configured from the following aspects: (1) Model structure optimization: Multi-task learning: Design the model to support multi-task learning, such as text classification, entity recognition, sentiment analysis, etc., to better understand and process news information. Attention mechanism: Introduce the attention mechanism to focus on the key information in the news text, which helps to generate more accurate lines. Hierarchical representation: Build a hierarchical model structure, first perform high-level summary and topic modeling on the news information, and then refine it to generate specific lines. (2) Role identification and line allocation; Role modeling: Build models for different podcast roles (such as host, guest, expert, interviewee) so that each role has a unique language style and expression. Line generation: Generate lines that match the role identity based on the role characteristics and news content. (3) Fine-tuning training: Dataset construction: Collect and annotate a large amount of news information and corresponding podcast line scripts to build a special training dataset. Use the deep learning framework PyTorch to train the model on the training set and adjust the hyperparameters to prevent overfitting. Domain Adaptation: By fine-tuning the model on data from a specific domain, the model is better adapted to the task of generating podcast scripts for news information. Continuous Learning: Implement a continuous learning mechanism to allow the model to continuously learn from new data over time to maintain its performance. For example, use the low-rank decomposition technology LoRA (Low-Rank Adaptation) for fine-tuning, and adjust the model's weight matrix by introducing two low-rank matrices to reduce the number of parameters and computational burden. Configure fine-tuning parameters, such as learning rate, batch size, and training cycle, to ensure model convergence. (4) Format Conversion: Template Engine: Develop a template engine to automatically organize the generated lines in the format of "role: lines". Natural Language Generation (NLG): Use NLG technology to ensure the natural fluency and semantic accuracy of the lines. (5) Interactivity and Participation: Dialogue Management: Optimize the model to generate interactive lines to simulate real dialogue scenarios. User Feedback Loop: Design a feedback mechanism to allow users to evaluate the generated scripts, and the model optimizes itself based on the feedback. (6) Sentiment and tone simulation: Sentiment analysis: Conduct sentiment analysis on news information to ensure that the lines can accurately convey the emotional color of the news; Tone modeling: Simulate different tones to make the expression of the lines richer and more vivid. (7) Multimodal fusion: Text and audio fusion: Fusion of text information with existing podcast audio clips to improve the realism and attractiveness of the lines. (8) Evaluation and optimization: Performance evaluation: Regularly evaluate the model, including the accuracy, naturalness and character consistency of the lines. Optimization algorithm: Adjust the model parameters and algorithms based on the evaluation results to improve the quality of script generation.

[0067] Exemplarily, the speech synthesis TTS unit is specifically used to convert the generated dialogue script into natural and fluent speech output, which may include the following configuration processes: (1) Front-end processing optimization: By introducing advanced models and algorithms, the pronunciation prediction accuracy of the text is improved. For example, deep learning technology can be used to understand and process the complex rules of natural language, including the pronunciation differences of polyphones, numbers, proper nouns, etc. in different contexts. (2) Back-end training model optimization: Optimize the parameters and structure of the model to improve the prediction ability of sound parameters. More complex network structures, such as convolutional neural networks (CNN) or recurrent neural networks (RNN), can be used to better learn and simulate the complex characteristics of speech. (3) Improvement of speech synthesis speed: By using methods such as the Fast Speech model to generate Mel spectrum in parallel, the speed of speech synthesis can be significantly improved while maintaining speech quality, so that the TTS system can respond more quickly and generate fluent and natural speech. (4) Improved interactivity and naturalness: By optimizing the algorithm, the TTS system can more naturally simulate the rhythm and intonation of human speech, including appropriate pauses, changes in intonation, and emotional expressions, making the simulated host and guest podcast more vivid and realistic. (5) Personalization and customization: Allow users to adjust the voice characteristics of the TTS system according to their personal preferences, such as speaking speed, pitch, volume, etc., so that the sound of each podcast is unique. (6) Fusion of speech and ambient sound: Optimize the TTS algorithm so that it can naturally integrate ambient sound into the speech output according to the context or user preference while synthesizing speech.

[0068] Furthermore, before calling the speech synthesis tts unit of the AI ​​model service module through the web_socket long connection service for speech synthesis, it is possible to determine again whether the user has new voice input. If there is new input, the retrieval operation is terminated and the operation of obtaining the user's interactive voice is returned to execute. If there is no new input, speech synthesis is performed to generate character audio, that is, podcast audio, and the podcast audio is fed back to the client through the web_socket long connection service to instruct the client to play the podcast audio to the user, thereby realizing feedback on the character: dialogue content.

[0069] Optionally, an interface for users to interact with the system can be provided on the client side, including voice input and audio output, using web-socket to provide real-time interruption capabilities, and based on a personalized recommendation algorithm, recommending relevant content based on user preferences and historical behavior to enhance the user experience.

[0070] Optionally, the intelligent podcast system can obtain the latest news or information in real time and quickly generate podcast content. Furthermore, after playing the target podcast script, it can re-retrieve content based on the user's new voice input, combined with the original topic and new questions, and quickly generate new podcast content.

[0071] Exemplarily, the implementation method of the intelligent podcast system can be as follows: the user inputs voice commands or questions through the voice recognition module. After receiving the voice input, the system converts it into text through the ASR module. The content generation module retrieves real-time news information based on the text content; combines the news information and the user input questions, and uses an optimized large model to convert the information into a podcast script such as (host: xxxx, guest: xxxx); the TTS module converts the script into voice output and plays it to the user through an interactive interface. Users can input voice commands at any time, such as changing topics and asking in-depth questions; the intelligent podcast system can adjust the content generation strategy through personalized recommendation algorithms based on user feedback and behavior, and can also monitor news sources or other information sources in real time to ensure timely updating and accuracy of content.

[0072] Exemplarily, based on the solution of the present invention, the interaction process between the intelligent podcast system and the user can be demonstrated by the following example:

[0073] User: How come the Nobel Prize in Physics was awarded to an AI expert?

[0074] Host A (search waiting): I think the topic you chose is also very interesting. Let me help you find it. Please wait a moment.

[0075] Host A (Opening) Hello everyone, welcome to the Tech Chat program. I am your host A. Today we are going to discuss a controversial topic, that is, this year's Nobel Prize in Physics was awarded to a scientist working in the field of artificial intelligence. This decision has caused widespread discussion in academia and the technology circle. Some people support it and believe that AI has made a great contribution to the development of physics, while others are skeptical and feel that this is not in line with the original intention of the Physics Prize. So, today I invited two very thoughtful guests to talk to us about this topic. They are Guest 1 and Guest 2. Hello, both guests, say hello to everyone.

[0076] Guest 1: xxx; Guest 2: xxx; (The two guests introduced and elaborated their views respectively)

[0077] User (interrupting to ask a deep question): Who was awarded this award last time?

[0078] Host (A) (searching and waiting): OK, I see. I'm preparing it for you. It will be ready soon. Host (A): Last year, the Nobel Prize in Physics was awarded to Pierre Agostini, Ferenc Krauss and Anne Lullier for their contributions to attosecond light pulses.

[0079] Guest 1: xxx; Guest 2: xxx; (The two guests elaborated their views respectively);

[0080] Moderator (A) (ending): Well, what you two said makes a lot of sense. Although interdisciplinary integration makes scientific research more efficient and diversified, how to maintain the independence and essential characteristics of each discipline is indeed a question we need to think about. Well, today's topic on the Nobel Prize in Physics ends here. Thank you to guests 1 and 2 for participating, and see you next time!

[0081] The technical solution of the present invention provides a new interactive intelligent podcast system, which can provide personalized and instant podcast content, enhance the user's interactive experience, and improve the efficiency and quality of information dissemination.

[0082] Embodiment 3

[0083] Figure 3 : is a structural block diagram of an interactive device based on a podcast system provided in Embodiment 3 of the present invention; this embodiment can be applicable to the situation where an intelligent podcast system responds to user needs and provides feedback to users in the form of podcast lines and scripts. The interactive device based on a podcast system provided in the embodiment of the present invention can execute the interactive method based on a podcast system provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method; the interactive device based on a podcast system can be implemented in the form of hardware and / or software, and configured in an electronic device having an interactive function based on a podcast system, such as an intelligent podcast system having an interactive function, such as Figure 3 As shown, the interactive device based on the podcast system specifically includes:

[0084] The recognition module 301 is used to recognize and process the interactive speech input by the user based on a preset automatic speech recognition algorithm during the interaction with the user to obtain target interactive information;

[0085] A matching module 302 is used to perform matching search in a preset knowledge base based on the original topic and the target interaction information according to the target interaction information, and obtain feedback information corresponding to the target interaction information when it is determined that the user has not changed the interaction topic;

[0086] The feedback module 303 is used to determine the target podcast role involved according to the feedback information, and determine the target podcast dialogue script corresponding to the feedback information based on the preset large language model according to the feedback information and the target podcast role, so as to provide feedback to the user.

[0087] The technical solution of the embodiment of the present invention is as follows: in the process of interacting with the user, based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain target interaction information; based on the target interaction information, when it is determined that the user has not changed the interactive topic, a matching search is performed in a preset knowledge base based on the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information; based on the feedback information, the target podcast role involved is determined, and based on the feedback information and the target podcast role, the target podcast script corresponding to the feedback information is determined based on a preset large language model to provide feedback to the user. By combining speech recognition, large model text generation and speech synthesis technology, personalized content can be better fed back to the user, the interactivity between the podcast system and the user can be increased, real-time interaction between the user and the podcast system can be achieved, and the user experience can be improved.

[0088] Furthermore, the recognition module 301 is specifically used to: interact with the user based on a preset communication protocol, and in the process of interacting with the user, obtain the interactive voice input by the user;

[0089] Based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain target interactive information; wherein the target interactive information is voice instructions or question information.

[0090] Furthermore, the matching module 302 is specifically used for:

[0091] Determine the original topic currently interacting with the user, and perform similarity matching between the target interaction information and the original topic;

[0092] If the similarity between the target interaction information and the original topic is greater than a preset similarity threshold, it is determined that the similarity match is successful, and it is determined that the user has not changed the interaction topic.

[0093] Furthermore, the matching module 302 is also used for:

[0094] According to the original topic, a preliminary screening is performed in the preset knowledge base to obtain the target knowledge content associated with the original topic;

[0095] According to the target interactive information, a matching search is performed in the target knowledge content, and feedback information corresponding to the target interactive information is generated according to the matching result; wherein the feedback information is news information or other knowledge information corresponding to the target interactive information.

[0096] Furthermore, the feedback module 303 is specifically used for:

[0097] Determine the language style and expression that corresponds to the target podcast persona;

[0098] According to the language style, expression method and feedback information, the preset large language model is used to convert the feedback information into the form of a podcast script, and the target podcast script corresponding to the feedback information is obtained.

[0099] Furthermore, the feedback module 303 is also used for:

[0100] Based on a preset speech synthesis algorithm, the podcast audio corresponding to the target podcast script is generated according to the target podcast script and the speech materials corresponding to the target podcast role;

[0101] Based on the preset communication protocol, the podcast audio is fed back to the user to complete the interaction process with the user.

[0102] Embodiment 4

[0103] Figure 4 It is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention. Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0104] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0105] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0106] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as an interactive method based on a podcast system.

[0107] In some embodiments, the interactive method based on the podcast system can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the interactive method based on the podcast system described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the interactive method based on the podcast system in any other appropriate manner (for example, by means of firmware).

[0108] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0109] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0110] In the context of the present invention, a computer readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer readable storage medium may be a machine readable signal medium. A more specific example of a machine readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0111] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0112] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0113] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0114] In one embodiment, the embodiment of the present invention also includes a computer program product, the computer program product includes a computer program, and when the computer program is executed by a processor, it implements the interactive method based on the podcast system of any embodiment of the present invention.

[0115] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present invention, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0116] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0117] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An interactive method based on a podcast system, characterized in that: include: In the process of interacting with the user, based on the preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain the target interactive information; According to the target interaction information, when it is determined that the user has not changed the interaction topic, a matching search is performed in a preset knowledge base according to the original topic and the target interaction information to obtain feedback information corresponding to the target interaction information; The target podcast role involved is determined according to the feedback information, and based on the feedback information and the target podcast role and based on a preset large language model, the target podcast dialogue script corresponding to the feedback information is determined to provide feedback to the user.

2. The method according to claim 1, characterized in that In the process of interacting with the user, based on the preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain the target interaction information, including: Interact with the user based on a preset communication protocol, and obtain interactive voice input by the user during the interaction with the user; Based on a preset automatic speech recognition algorithm, the interactive speech input by the user is recognized and processed to obtain target interactive information; wherein the target interactive information is voice instructions or question information.

3. The method according to claim 1, characterized in that Determine whether the user changes the interaction topic, including: Determine the original topic currently interacting with the user, and perform similarity matching between the target interaction information and the original topic; If the similarity between the target interaction information and the original topic is greater than a preset similarity threshold, it is determined that the similarity match is successful, and it is determined that the user has not changed the interaction topic.

4. The method according to claim 1, characterized in that: According to the original topic and the target interaction information, a matching search is performed in the preset knowledge base to obtain feedback information corresponding to the target interaction information, including: According to the original topic, a preliminary screening is performed in the preset knowledge base to obtain the target knowledge content associated with the original topic; According to the target interactive information, a matching search is performed in the target knowledge content, and feedback information corresponding to the target interactive information is generated according to the matching result; wherein the feedback information is news information or other knowledge information corresponding to the target interactive information.

5. The method according to claim 1, characterized in that According to the feedback information and the target podcast role, based on the preset large language model, the target podcast script corresponding to the feedback information is determined, including: Determine the language style and expression that corresponds to the target podcast persona; According to the language style, expression method and feedback information, the preset large language model is used to convert the feedback information into the form of a podcast script, and the target podcast script corresponding to the feedback information is obtained.

6. The method according to claim 1, characterized in that Provide feedback to users, including: Based on a preset speech synthesis algorithm, the podcast audio corresponding to the target podcast script is generated according to the target podcast script and the speech materials corresponding to the target podcast role; Based on the preset communication protocol, the podcast audio is fed back to the user to complete the interaction process with the user.

7. An interactive device based on a podcast system, characterized in that: include: The recognition module is used to recognize and process the interactive speech input by the user based on a preset automatic speech recognition algorithm during the interaction with the user to obtain the target interactive information; A matching module is used to perform matching retrieval in a preset knowledge base based on the original topic and the target interaction information according to the target interaction information, and obtain feedback information corresponding to the target interaction information when it is determined that the user has not changed the interaction topic; The feedback module is used to determine the target podcast role involved according to the feedback information, and determine the target podcast script corresponding to the feedback information based on the preset large language model according to the feedback information and the target podcast role, so as to provide feedback to the user.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executed by the at least one processor, and the computer program is executed by the at least one processor so as to enable the at least one processor to execute the podcast system-based interaction method described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the interactive method based on a podcast system according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, which, when executed by a processor, implements the podcast system-based interaction method according to any one of claims 1-6.

Images