90 results about "Timed text" patented technology

The invention belongs to the technical field of natural language processing, and discloses a text data labeling method and system based on weak supervised learning. The method comprises the following steps: performing pseudo label generation on a plurality of historical text data by using a weak supervised learning algorithm to obtain a plurality of source domain data provided with real labels and a plurality of target domain data provided with pseudo labels; constructing a text data labeling model by using a deep learning and adversarial training algorithm, and adjusting the text data labeling model by using a continuous learning algorithm; and labeling the real-time text data by using the adjusted text data labeling model to obtain labeled real-time text data, and updating the adjusted text data labeling model by using a continuous learning algorithm. The problems that in the prior art, manual labeling cost is high, labeling quality is uneven, and dependence on label data is high are solved.

The invention discloses an AI digital human conference proxy method and device under an offline local area network and a medium, and the method comprises the steps: collecting conference voice in real time, converting the conference voice into a real-time text, and pre-judging a subject set in combination with a localized industry knowledge base and a user historical conference track; if it is detected that the user or the to-be-decided item is mentioned, reply voice is generated in combination with historical corpora of the user; and if the conference enters the pre-judgment topic set, calling the pre-loaded user feature packet to generate reply voice, and controlling the digital person to generate a corresponding audio and video stream. The invention provides an AI digital human conference proxy method and device under an offline local area network and a medium, and aims to realize topic pre-judgment in combination with a local knowledge base and a historical conference track of a user and provide preparation for real-time reply; meanwhile, for different scenes, the user historical corpus and the user feature packet are called respectively to generate the reply, so that the problem that the real-time personalized reply generation of the AI digital person conference agency and the conference issue pre-judgment are difficult to collaboratively realize in an offline local area network environment can be solved.

The invention relates to a public opinion information intelligent processing method and system based on multi-modal data fusion. The method comprises the steps of obtaining real-time text data, image data, video data and social attribute data; performing keyword rule matching and deep learning model analysis on the text data in parallel, comparing confidence coefficient difference values of rule output and model output, and determining a text analysis result; performing cross-modal fusion on the multi-source feature vectors, and dynamically distributing feature weights according to the data quality of each modal to generate weighted fusion features; calculating a social influence value based on the user authority, the propagation path depth and the time attenuation coefficient in the social attribute data, and identifying a key community cluster; further carrying out weighted matching on a pre-constructed domain strategy matrix through multi-modal evidence; and adjusting a rule weight and a model parameter according to the target domain configuration result and the weighted fusion feature. According to the method, the public opinion field can be accurately identified, the processing scheme is intelligently matched, and a visual and understandable decision basis is provided.

Systems and methods for improved video captions are disclosed. The methods for providing timed text comprise obtaining a plurality of subtitles for a media asset; determining a playback rate of the media asset; determining a relevance index based on the playback rate; and selecting a set of the plurality of subtitles based on the relevance index. The methods may further comprise determining an indication of priority of subtitles within the set of subtitles, rendering the set of subtitles according to the indication of priority, wherein the highest priority subtitles are rendered first.

The invention belongs to the technical field of call service processing, and discloses a real-time call question and answer matching processing method, which comprises the following steps of: converting voice into characters in real time, and sending preprocessed voice data into an ASR service for real-time conversion; real-time text processing and grouping are carried out, and texts are extracted, sorted and grouped through role detection and grouping and timestamp correction; and processing the text after the call, and matching the text with the identified groups in the call process. According to the method, calling of the TEXTCNN model is dispersed to the call process, so that the processing method remarkably reduces the time required for centralized text processing after the call is ended, which means that even if the call duration is relatively long, the response duration of filing information can be effectively controlled due to the fact that the processing is dispersed to the whole call period, and the user experience is improved. Therefore, the performance requirement of outputting filing information within 5 seconds after on-hook can be met even in peak hours.

The invention discloses a method and system for realizing real-time two-way visual interaction of a digital human by calling a camera, and belongs to the technical field of multi-modal interaction, and the method comprises the steps: collecting video streams of a user and an environment in real time through a mobile terminal camera, and synchronously obtaining a natural language question of the user; screening related video keys based on question semantics; performing cross-modal semantic fusion analysis on the question text and the key frame image by using a multi-modal model to generate a corresponding answer result; and the virtual digital person outputs an answer through voice synthesis and synchronous anthropomorphic animation, so that real-time image-text question and answer interaction between the digital person and the user is realized. Aiming at the problem that the existing digital human interaction lacks visual situation feeling, the invention provides an end-cloud collaborative visual question-answer interaction scheme, so that the video semantic processing load of a mobile end can be effectively reduced, the end-cloud collaborative real-time image-text question-answer interaction is realized, the perception ability of the digital human to the environment is greatly improved, and the user experience is improved. And the intuition and the naturalness of user interaction experience are improved.

Systems, methods, and apparatuses disclosed herein can stream a live session in real-time, or near real-time, over the Internet. The live session can include live streaming content, such as audio content, video content, live closed captioning, ticker tape, data and graphical overlays, and / or real-time text, among others, that is being presented by a live host. These systems, methods, and apparatuses can invite one or more persons to participate in the live streaming content, also referred to as live participants, to enter a virtual lobby that is associated with the live session to participate in the live streaming content with the live host. The virtual lobby represents a designated virtual staging area, for example, a virtual space, that provides a controlled and organized environment to enable the one or more live participants to gather and to prepare for the live session. These systems, methods, and apparatuses can virtually move the live host and / or the one or more live participants between the live session and the virtual lobby. For example, these systems, methods, and apparatuses can virtually move the one or more live participants from the live session to the virtual lobby to leave the live session and no longer participate in the live streaming content. In this example, these systems, methods, and apparatuses can virtually move the one or more live participants to the live session from the virtual lobby to join the live session once again to participate in the live streaming content.

The invention discloses an information processing method and device, and the method comprises the steps: responding to a condition meeting a first condition, and outputting first multimedia information; the first multimedia information comprises first text information; wherein the first multimedia information is generated based on first information, the first information at least comprises the first text information, and the first text information and first time information can form a first time text relation.

This invention discloses a long text matching method combining noise filtering and a divide-and-conquer strategy. The method includes: constructing a long text matching model, which comprises a keyword extraction layer, an association extraction layer, and a filtering layer. The keyword extraction layer extracts keywords from the text; the filtering layer filters noise from the text based on sentence similarity to obtain a denoised text sequence; and the association extraction layer further removes keywords from the denoised text sequence to obtain the remaining associated text. The long text matching model is trained with the optimization objective of minimizing a set overall loss function, which reflects the global matching distribution and combines the keyword and association matching distributions. For the target text, real-time text matching is performed using the trained long text matching model. This invention improves the generalization ability and accuracy of text matching.

A computer-implemented method comprising: obtaining real-time text data relating to a matter, the text data comprising a plurality of portions of information; performing a categorisation process, wherein the categorisation process is configured to run a plurality of threads in parallel, wherein each thread of the plurality of threads acts on one portion of information at a time, wherein each thread performs the following steps: (i) obtaining a sentiment score based on the portion of information using a Sentiment Analysis machine learning, ML, model; (ii) assigning a category to the matter based on the sentiment score using a classification ML model trained on historical data; and (iii) updating a live category based on the category assigned to the matter; and finally, outputting the live category to a user in real-time.

The application relates to the technical field of natural language processing, financial technology and medical technology, and discloses a standard time conversion method and device based on natural language, equipment and a medium, which comprises the following steps: collecting current session text information in real time; when target text representing a time attribute appears in the text, acquiring corresponding collection time and standardizing the collection time into a specified format reference time; extracting preset time period text containing the target text as context information; reasoning the target text and the context by combining the reference time with a preset natural language model to determine a target date, a target time word and a target number of days; and performing consistency verification on the target time word and the target number of days, and outputting the target date if the verification is passed. The application can be applied to a data processing platform in the fields of financial technology and medical health, improves the standardization conversion efficiency of time text strings, improves the accuracy of large models, and improves the convenience of the corresponding platform.

The present invention belongs to the technical field of natural language processing, and discloses a text data annotation method and system based on weakly supervised learning. The method comprises the following steps: using a weakly supervised learning algorithm to generate pseudo labels for a number of historical text data, and obtaining a number of source domain data with real labels and a number of target domain data with pseudo labels; using a deep learning and adversarial training algorithm to construct a text data annotation model, and using a continuous learning algorithm to adjust the text data annotation model; using the adjusted text data annotation model to annotate real-time text data, and obtaining annotated real-time text data, and using a continuous learning algorithm to update the adjusted text data annotation model. The present invention solves the problems of high manual annotation cost, uneven annotation quality, and high dependence on label data in the prior art.

The invention discloses a device control method based on artificial intelligence, and belongs to the technical field of data processing, and the method comprises the following steps: S1, obtaining a real-time voice instruction of a scene, and converting the real-time voice instruction into a real-time text instruction; s2, matching an optimal preset instruction for the real-time text instruction in the instruction set according to a real-time instruction flag function of the real-time text instruction; and S3, determining equipment corresponding to the optimal preset instruction. According to the invention, the voice instruction appearing in the scene can be comprehensively analyzed to obtain the most accurate preset instruction, and the equipment is controlled through the preset instruction, so that accurate control of various equipment can be realized, and the intelligent degree of the equipment is improved.

This invention belongs to the field of text printing technology and discloses a text printing method and system based on speech recognition. The method includes the following steps: constructing a text printing template database; constructing a mixed-language speech recognition model; acquiring speech audio data in real time and performing speech recognition; matching several corresponding text printing templates; selecting a text printing template; fusing the real-time speech text data with the selected text printing template; and printing the real-time text printing data. The system includes a database construction unit, a model construction unit, a storage unit, a speech audio acquisition unit, a speech recognition application unit, a template matching unit, a template selection unit, a data fusion unit, and a printer. This invention solves the problems of low intelligence, poor speech recognition effect, low recognition efficiency, and lack of organic integration in existing technologies.

The invention discloses a portable real-time text recognition and translation method and system based on a lightweight Transform, and belongs to the technical field of computer vision and natural language processing. The system comprises an image preprocessing module, a text recognition module, a text translation module, a result presentation module, a model optimization engine and a system control module. A lightweight Transform architecture is adopted, and the calculation complexity is reduced through a layered mixed attention mechanism, a shared weight design, dynamic depth adjustment and a compression feedforward network. The text recognition module adopts a CNN and Transform hybrid architecture to realize detection and recognition, and the text translation module shares partial encoder weight and supports incremental decoding. The system integrates compression technologies such as knowledge distillation, quantification and pruning, realizes end-side off-line real-time operation in combination with hardware perception optimization, supports multi-language image-text translation, and has the characteristics of high precision, low delay and low power consumption.

This invention relates to the field of artificial intelligence technology, and more particularly to a method, apparatus, computer device, and medium for generating incorrect sentences for model training. The method maps the number of terms in the text to be processed to an adjustment quantity, samples target terms from the text to be processed, adjusts the target terms to obtain augmented text, occludes the text to be processed, inputs it into a reconstruction model to obtain reconstructed text, trains a generative model based on the text to be processed, the augmented text, and the reconstructed text, inputs real-time text into the trained generative model to obtain generated incorrect sentence text, processes the text to be processed using multiple data augmentation methods to obtain text containing rich error types as labels, trains an end-to-end generative model, allowing real-time text to be directly input into the trained generative model to obtain generated incorrect sentence text without frequent switching of data augmentation methods, greatly improving the efficiency of incorrect sentence generation while ensuring high quality.

The invention discloses a text image quality detection method, system and device and a medium, and the method comprises the steps: obtaining a text image, and dividing the text image into a training set and a test set; constructing a text image quality detection model, and inputting the training set into the text image quality detection model for training; and inputting the test set into the trained text image quality detection model to obtain a detection result, and evaluating and optimizing the text image quality detection model. According to the method, the text image quality detection model without the reference image is adopted to detect the text image quality, a fixed reference image is not needed, the method is more suitable for an actual application scene and is more flexible and practical, the detection precision of text image quality evaluation is improved, and the accuracy and robustness of text image quality detection are improved; and accurate and real-time text image quality evaluation is realized.

This specification provides a dialogue analysis system and method. The system includes: an acquisition module for acquiring a text sequence, wherein the text sequence is obtained by real-time text conversion of classroom dialogues; a matching module for searching, in a pre-established rule base, using a large language model, whether a target sequence matching the text sequence exists; wherein the rule base includes classification rule sequences for various dialogue types and / or reference text sequences corresponding to each dialogue type; and an analysis module for analyzing, based on the target sequence matched by the matching module, to obtain the target dialogue type corresponding to the classroom dialogue. This disclosure combines a rule base with a large language model to achieve usability and flexibility in classroom dialogue analysis, ensuring both deep integration of the dialogue analysis system with educational theory and the ability to process complex dialogue sequences through a large language model, achieving accurate dialogue classification and analysis.

The invention belongs to the technical field of natural language processing, and discloses a text generation model construction method and system based on artificial intelligence. The method comprises the following steps: constructing an initial integrated text generation model, and setting an experience playback pool and an elastic loss function; performing optimization training on the initial integrated text generation model according to the training text data to obtain a trained integrated text generation model and a plurality of historical training experiences, and storing the trained integrated text generation model and the historical training experiences in an experience playback pool; extracting a plurality of real-time text generation experiences, and randomly extracting a plurality of historical training experiences from the experience playback pool; and based on the elastic loss function, according to a plurality of historical training experiences and a plurality of real-time text generation experiences, continuously training the trained integrated text generation model to obtain an updated integrated text generation model. According to the method, the problems of unstable generation quality, strong data dependence, model function overflow and insufficient generalization ability in the prior art are solved.

Emergency calls that use real time text (RTT) in addition to audio face a challenge: between an originating endpoint (e.g., a cellular network) sending a session initiation protocol (SIP) invite and receiving the response (e.g., 200 OK) from a downstream terminating endpoint (e.g., a peering network), the media attribute fields associated with the text and audio data may become swapped. Three options are disclosed for dealing with this issue. (1) The originating endpoint may swap the protocol paths for the RTT and audio data so that the public safety answering point (PSAP) and the user equipment (UE) are able to communicate. (2) The originating endpoint may send a correction request to the terminating endpoint. (3) If default protocol paths are specified for RTT and audio, the originating endpoint uses the default protocol paths independently of whether the SIP invite or response specified the correct data type and protocol path association.

The application discloses a kind of text image quality detection method, system, equipment and medium, the method includes: obtaining text image, text image is divided into training set and test set;Text image quality detection model is constructed, and training set is input into text image quality detection model for training;Test set is input into the text image quality detection model that has been trained, obtains detection result, and text image quality detection model is evaluated and optimized.The application detects the quality of text image using the text image quality detection model without reference image, does not need fixed reference image, is more suitable for practical application scene, is more flexible and practical, improves the detection precision of text image quality evaluation, and improves the accuracy and robustness of text image quality detection, realizes accurate and real-time text image quality evaluation.

A scalable, automated timed text workflow system designed to optimize the generation and refinement of time-synchronized textual content for video is disclosed. Integrated Machine Translation (MT) models and advanced AI-driven tools such as Computer Vision, Generative AI, and Traditional AI automatically generate and refine timed text, including subtitles, closed captions (CC), and SDH (Subtitles for the Deaf and Hard of Hearing). Human review is coordinated through a workflow orchestration system when needed, offering flexibility and scalability for handling high-volume media processing across various platforms and formats.

The present invention discloses a method and system for dynamically identifying abnormal SMS sending based on big data technology, the method comprising: collecting historical SMS data, and performing aggregate statistics to obtain offline statistical results; performing statistics on real-time SMS data, and merging the real-time statistical results with the offline statistical results; generating an intelligent judgment threshold according to the offline statistical results; judging whether the real-time statistical results meet the requirements according to the judgment threshold, and issuing an early warning for the real-time statistical results that do not meet the requirements. By collecting historical SMS data, feature processing can be performed on the date frequency, the corresponding sending data can be obtained, and fitting training can be performed according to the features, so as to construct a trend prediction model that can predict the number of sendings of different date frequencies, so as to output the predicted number of SMS sendings, and at the same time, the judgment threshold can be calculated in conjunction with the set artificial rule threshold, so as to dynamically identify abnormal SMS sending behavior and accurately and timely issue an early warning.

The embodiment of the invention provides a real-time data processing method and device and electronic equipment, and the real-time data processing method comprises the steps: building a session with a transmitting end, and obtaining real-time text data which is transmitted by the transmitting end and is related to an operation event; determining whether the operation events comprise risk operation events or not according to the real-time text data; and when the operation event comprises a risk operation event, stopping the operation event or terminating the session.

The invention provides a dialogue analysis system and method, and the system comprises an obtaining module which is used for obtaining a text sequence; wherein the text sequence is obtained by performing real-time text conversion on classroom dialogues; the matching module is used for searching whether a target sequence matched with the text sequence exists or not through a large language model in a pre-established rule base; wherein the rule base comprises classification rule sequences of multiple dialogue types and / or reference text sequences corresponding to each dialogue type; and the analysis module is used for performing analysis according to the target sequence matched by the matching module to obtain a target dialogue type corresponding to the classroom dialogue. According to the method, the rule base and the large language model can be combined, usability and flexibility of classroom dialogue analysis are achieved, deep integration of the dialogue analysis system on the education theory is ensured, complex dialogue sequences can be processed through the large language model, and accurate dialogue classification and analysis are achieved.

The application discloses a network information collection method and system based on real-time text analysis, comprising the following steps: acquiring real-time text data, preprocessing the real-time text data, acquiring first data and second data according to the preprocessed real-time text data, calculating the similarity of the first data and the similarity of the second data, weighting the similarity of the first data and the similarity of the second data to obtain a classification target, constructing a text classification model according to the classification target, inputting the real-time text data into the text classification model to obtain classification data, and outputting the classification data as collected network information. The method can not only improve the accuracy of network information collection, but also has good interpretability and can be directly applied to a network information collection system.

The present application relates to a memory-enhanced deep personalized input method system based on end-side large language model, and belongs to the technical field of artificial intelligence. The present application solves the problems of the prior art, such as the loose coupling of the model and the input core process, the shallow personalization mechanism, the delay and privacy risk caused by cloud deployment, and the insufficient efficiency of the end-side operation that cannot meet the real-time input requirements. The system includes a front-end real-time interaction module, a stylized large model post-training module, a hierarchical memory mechanism module, and an end-side reasoning optimization module. The present application can realize low-delay real-time text generation under the condition of highly limited mobile terminal resources, and at the same time, through continuous modeling of user input behavior, it can complete deep personalization adaptation and ensure the end-side retention and privacy security of user input data throughout the process.

This invention provides a method for determining the placement location and type of a centralizer based on multi-source data fusion, comprising: Step 1, acquiring casing record table data before the casing operation begins; Step 2, acquiring real-time text data and real-time video data at the start of the casing operation; Step 3, determining the target casing information currently being lowered and the type of centralizer to be placed therein based on the real-time text data; Step 4, determining the type of centralizer actually placed on the target casing based on the real-time video data; Step 5, determining the specific time of centralizer placement on the target casing based on the real-time video data; Step 6, determining whether the corresponding type of centralizer was correctly placed on the target casing during the actual placement process. This method for determining the placement location and type of a centralizer based on multi-source data fusion standardizes the casing lowering operation process, effectively replaces traditional manual monitoring methods, and improves on-site safety and supervision efficiency.

The techniques disclosed herein provide enhanced controls for the display of real-time text (RTT) in calls and meetings. RTT is the ability for someone to send a text message on a character-by-character basis to everybody else in a call or meeting. The system disclosed herein integrates RTT, video, and live captions all in one central experience. This integrated experience allows users to participate equitably by making RTT accessible to users regardless of the operating mode they are in and still concurrently access other meeting content, including video streams, chat messages, live captions, transcripts, and artificial intelligence (AI) tools, such as Copilot. In one embodiment, during an online conference, in response to one of the attendees activating a RTT mode, when at least one user minimizes a meeting stage, such as for the purpose of multitasking while listening, the conference application maintains a display area for displaying RTT.

The invention discloses a real-time online text violation recognition method and system and a storage medium, and belongs to the technical field of text recognition. The method comprises the following steps: collecting user behavior data in real time, and generating a user-level weak tag; comparing and learning the user-level weak tags to obtain a single sentence code and a single sentence risk assessment model; automatically distributing a pseudo label and an attention weight for each sentence by utilizing a pseudo label distribution rule, and training to obtain a lightweight sentence-level model; for an input real-time text, calculating a violation probability through a lightweight sentence-level model to obtain a risk index; according to the risk indexes and a preset risk threshold value, performing grading processing, blocking users whose risk indexes are in a high risk interval, performing secondary reasoning and random sampling on the users whose risk indexes are in a middle and low risk area, and then performing manual auditing; and iteratively updating the user-level weak label according to a user disposal result and newly added report data, and realizing self-evolution of each model. A large number of user-level tags can be obtained at low cost, and the detection rate is improved.