STT Result Linking with Reference Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of speech-to-text (STT) services face difficulties in identifying important parts from the vast amount of data generated, including STT result text, notes, and related content, as it is challenging to determine which parts are crucial among the diverse data types.
Innovation Solution
A method is provided to determine important parts within STT results and reference data by acquiring STT data and reference data, identifying first important information based on user input or predetermined keyword information, and linking this information with similar parts in other data using word similarity, sentence embedding vectors, or question answering models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all diverse data types (STT result text, notes, related content) are provided to users, then information completeness is improved, but information overload and difficulty in identifying important parts increases
Solution Approach 1:
The patent extracts important parts from diverse data types (STT result text, notes, related content) by comparing them against reference data. The system identifies and extracts key information that matches reference data, separating important content from unnecessary content. This extraction process resolves the contradiction by providing only relevant information to users, maintaining information completeness while eliminating overload.
Solution Approach 2:
The patent segments diverse data types into distinct categories (STT result text, notes, related content) and processes each separately. By dividing the data processing into segments and applying reference data comparison to each segment, the system identifies important parts within each category. This segmentation approach allows users to view organized, categorized important information rather than a overwhelming mix of all data types.
2Ease of operation
If important parts are extracted from diverse data types, then ease of viewing important information is improved, but the complexity of determining which parts are important increases
Solution Approach 1:
The patent introduces reference data as an intermediary to objectively determine important parts. Instead of relying on complex user preferences or subjective importance criteria, the system uses reference data (such as transcripts, summaries, or external knowledge bases) as a mediator to identify which parts of diverse data types are important. This intermediary approach simplifies the determination process by providing an objective basis for importance assessment.
Solution Approach 2:
The patent employs feedback mechanisms where the system continuously compares extracted important parts against reference data and refines the extraction process. By using feedback from the comparison results, the system learns and improves its ability to identify important parts across different data types. This feedback loop reduces complexity by automating the determination process rather than requiring manual configuration of complex importance criteria.
3Measurement precision
If reference data comparison is performed to identify important parts, then accuracy in identifying important information is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by performing reference data comparison only on segments of data that are likely to contain important information, rather than comparing every single part of all diverse data types. The system identifies candidate important parts through preliminary analysis and then performs detailed reference data comparison only on these candidates. This partial approach maintains high accuracy in identifying important information while significantly reducing overall processing time compared to exhaustive comparison.
Data Source
AI summary
Disclosed is a method for determining important parts among a speech-to-text (STT) result and reference data, which is performed by a computing device. The method may include acquiring STT data generated by performing STT with respect to a speech signal; acquiring reference data; determining first important information in one data among the STT data and the reference data; and determining second important information linked with the first important information in other data different from data in which the first important information is determined among the STT data and the reference data.