Multimodal Text Enhancement for Personalized Speech-to-Text
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech recognition technologies often require manual corrections for grammatical errors and personalization of speech-to-text messages, providing inaccurate grammar solutions and lacking personalized content insertion.
Innovation Solution
A method and system that utilize a natural language processing model to determine features of text blocks, identify usage patterns through a pattern identification neural network, and enhance multimodal inputs by generating personalized text with emoticons, emojis, and lingo, eliminating the need for manual corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If speech recognition technology is used to convert spoken language to text, then the speed of text input is improved, but the text contains grammatical errors and lacks punctuation marks requiring manual correction
Solution Approach 1:
The patent introduces an intermediary processing system that includes text block segmentation, feature determination (context, gender, lingo usage), and pattern identification neural networks. This intermediary layer processes the raw speech-to-text output before final delivery, automatically correcting grammatical errors and adding punctuation without requiring manual intervention.
Solution Approach 2:
The system performs preliminary actions by pre-processing the speech-to-text content through multiple neural network models that identify usage patterns from historic data. Grammar corrections, punctuation additions, and personalization (emoticons, emojis) are applied in advance before the text is presented to the user, eliminating the need for post-processing manual corrections.
2Manufacturing precision
If manual corrections are provided for speech-to-text messages, then text accuracy is improved, but time consumption increases
Solution Approach 1:
The system implements self-service by enabling automatic self-correction of speech-to-text messages. The neural network models analyze the input text, identify errors, and perform corrections autonomously using learned patterns from historic user data. The system also automatically personalizes messages by adding appropriate emoticons, emojis, and lingo without user intervention, making the correction process entirely self-service oriented.
3Manufacturing precision
If existing grammar correction systems are used on continuous text, then some grammatical errors are corrected, but the correction accuracy is insufficient
Solution Approach 1:
The patent segments the continuous speech-to-text blob into discrete text blocks based on detected text boundaries. Each text block is independently processed through the feature determination module and pattern identification neural network. This segmentation allows the system to apply context-specific corrections to each block, improving overall grammar correction accuracy while managing system complexity through modular processing.
Solution Approach 2:
The system changes parameters by determining multiple features (context, gender information, contextual lingo information, lingo usage information) for each text block. These parameter changes enable the neural network to adapt corrections based on user-specific patterns and contextual factors, significantly improving grammar correction accuracy beyond traditional rule-based systems.
4Speed
If speech-to-text conversion is performed without personalization, then processing speed is maintained, but user experience and message readability are reduced
Solution Approach 1:
The system performs preliminary personalization actions by automatically adding emoticons, emojis, and user-specific lingo to speech-to-text messages before delivery. Using historic user data and pattern identification, the system pre-customizes messages to match individual user communication styles, improving readability and user experience without adding manual personalization steps that would slow down processing.
Data Source
AI summary
Present disclosure relates to a method and a user equipment (UE) for enhancing multimodal input content. The UE obtains text-blocks associated with multimodal input and determines a plurality of features for each of the text-blocks using a natural language processing (NLP) model and a neural network (NN) model. Further, the UE identifies usage pattern of the plurality of features based on historic data associated with the usage of the plurality of features using NN model. Thereafter, the UE generates personalized text for each of the textblocks based on the plurality of features and usage pattern. The UE inserts personalized emoticons based on context of text-blocks and lingo of each word using user lingo database. Thus, the UE eliminates manual correction of texts and enhances the readability of message/text entered by user.