Vocabulary Entry Generation for Out-of-Vocabulary Words
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech processing systems are unable to automatically generate vocabulary entries for out-of-vocabulary (OOV) words, especially special words with unique morphology, such as foreign words, family names, and abbreviations, leading to cumbersome and time-consuming correction and addition processes.
Innovation Solution
A method and device for automatically generating vocabulary entries from input acoustic data using vocabulary entry type-specific acoustic phonetic transcription and classification, along with phoneme-to-grapheme conversion, which includes neural network-based language identification and phonetic distance measures to classify and convert phonetic transcriptions into grapheme forms for various vocabulary types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If standard speech recognition systems are used, then they can process standard words fitting language morphology, but they cannot automatically detect or generate vocabulary entries for out-of-vocabulary words including special words with foreign morphology
Solution Approach 1:
The invention segments the vocabulary generation process into distinct phases: acoustic phonetic transcription of the input word, classification of the phonetic structure type, and application of type-specific grapheme conversion rules. This segmentation allows the system to handle diverse word types (standard words, foreign words, abbreviations, names) through specialized processing paths for each category, thereby achieving both automation and adaptability to special words
Solution Approach 2:
The system changes processing parameters based on the classified phonetic structure type. Different grapheme-to-phoneme conversion models are applied depending on whether the word is classified as a standard word, foreign word, abbreviation, or proper name. This parameter adaptation enables the automated system to generate accurate vocabulary entries for diverse word types that would otherwise require manual intervention
2Reliability
If manual correction and addition of OOV words is performed, then vocabulary entries can be added, but the process becomes cumbersome and time-consuming
Solution Approach 1:
The system performs self-service by automatically detecting out-of-vocabulary words, classifying their phonetic structures, and generating appropriate vocabulary entries without human intervention. The automated classification and grapheme conversion processes enable the system to add new words to the vocabulary autonomously, eliminating the time-consuming manual correction process while maintaining reliability through structured processing rules
3Productivity
If automatic vocabulary generation is implemented for all word types, then productivity increases, but system complexity increases due to multiple classification and conversion processes
Solution Approach 1:
The system manages complexity through segmentation of the processing pipeline into distinct modules: acoustic transcription module, classification module, and grapheme conversion module. Each module handles a specific aspect of vocabulary generation, making the overall complex system manageable and maintainable while achieving high productivity through automated parallel processing of multiple word types
Solution Approach 2:
The invention implements a universal vocabulary generation framework that handles multiple word types (standard words, foreign words, abbreviations, proper names) through a single integrated system. The classification module identifies different phonetic structure types, and the appropriate grapheme conversion model is automatically selected, allowing one system to serve multiple functions and achieve high productivity without requiring separate manual processes for each word type
Data Source
AI summary
A method and a device (1) for automatically generating vocabulary entry from input acoustic data (3), comprising a vocabulary entry type-specific acoustic phonetic transcription module (2; T) and a classifier module (6; 6′) for the classification of vocabulary entry types on the basis of the phonetic structure, wherein the classification of vocabulary entries is carried out in accordance with a number of predetermined types; and vocabulary entry type-specific phoneme-to-grapheme conversion means (28), to derive the respective vocabulary entries comprising a pair of a phonetic transcription and its grapheme form.