Speech-to-Text Word Identification Using Edit Distance Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice recognition technologies, such as speech-to-text, often produce imperfect text outputs that require manual correction, and text summarization methods face challenges in representing documents effectively, particularly in identifying and replacing similar words with precision.
Innovation Solution
A method for identifying words with similar pronunciations in text information by determining a target word and setting an edit distance threshold, allowing for batch changes in speech-to-text outputs and improving summarization by dynamically adjusting the threshold based on word length and using phonetic conversions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If voice recognition is used to convert speech to text, then speech can be processed in text format, but the output text is often imperfect and requires manual correction
Solution Approach 1:
The system automatically identifies and corrects similar words in STT output without requiring manual intervention. The word identification device performs self-service by detecting target words and their similar words, then batch-replacing them, allowing the system to correct its own output errors autonomously.
Solution Approach 2:
The patent replaces manual mechanical correction with an automated computational system. Instead of users manually checking and correcting each word, the system uses edit distance calculations and phonetic matching algorithms to automatically identify and replace similar words, substituting human labor with computational processing.
2Manufacturing precision
If users manually check and change words in STT output, then text accuracy can be improved, but it is time-consuming and cumbersome
Solution Approach 1:
The system performs automatic word identification and batch replacement without user intervention. The device calculates edit distances, identifies similar words based on phonetic conversions, and replaces them automatically, eliminating the need for users to manually check each word and significantly reducing correction time.
Solution Approach 2:
The system performs preliminary identification of target words and their similar words before the actual replacement operation. By pre-calculating edit distances and identifying candidates for replacement, the system prepares the correction data in advance, enabling efficient batch processing that reduces overall correction time.
3Productivity
If extraction summarization is used to create sentences from document words, then text can be generated efficiently, but selected words may not sufficiently represent the document
Solution Approach 1:
The patent changes the parameter for word selection from simple extraction to selection based on edit distance and phonetic similarity. By using these parameters to identify target words and their similar words, the system ensures that selected words more accurately represent the original document while maintaining generation efficiency through automated batch processing.
Data Source
AI summary
Disclosed is a method for identifying a word corresponding to a target word in text information, which is performed by one or more processors of a computing device. The method may include: determining a target word; determining a threshold for an edit distance associated with the target word; determining a word of which the edit distance from the target word among words included in text information is equal to or less than the threshold; and identifying the word corresponding to the target word based on the determined word.