Speech Recognition Phoneme Matching Personalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional speech recognition systems face accuracy issues when encountering specialist jargon, proper nouns, or words not included in their language models, leading to incorrect interpretations, particularly in domains like medicine or natural language processing.
Innovation Solution
A personalized speech recognition system that combines a general automatic speech recognition module with a personalized module, using phoneme generation and comparison techniques to identify and incorporate personalized language, such as contact names, into the text representation of audio inputs, thereby improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a general language model is used in speech recognition, then the system can process a broad range of general language, but accuracy deteriorates when encountering specialist jargon, proper nouns, or personalized language not contained in the model
Solution Approach 1:
The patent combines a general language model with a personalized language model into a unified speech recognition system. The general model handles broad language coverage while the personalized model supplements it with user-specific vocabulary, achieving both versatility and accuracy simultaneously
Solution Approach 2:
The system performs preliminary action by proactively collecting and storing personalized language data (contacts, messages, media metadata) in advance. This pre-collected personalized vocabulary is then integrated into the language model before speech recognition occurs, enabling accurate recognition of proper nouns and specialist terms
2Measurement precision
If the language model is expanded to include personalized words and phrases, then recognition accuracy for personalized language improves, but device complexity increases due to data collection and model management requirements
Solution Approach 1:
The system automatically collects personalized language data from existing device resources (contacts, messages, media files) without requiring manual user input. This self-service approach to building the personalized language model reduces operational complexity while maintaining high recognition accuracy
Solution Approach 2:
The speech recognition system is designed to handle multiple functions: general speech recognition, personalized language recognition, and automatic personalized vocabulary extraction from various device data sources. This multi-functional design consolidates complexity into a unified system rather than requiring separate systems for each function
Data Source
AI summary
A system, method and computer program is provided for generating customized text representations of audio commands. A first speech recognition module may be used for generating a first text representation of an audio command based on a general language grammar. A second speech recognition module may be used for generating a second text representation of the audio command, the second module including a custom language grammar that may include contacts for a particular user. Entity extraction is applied to the second text representation and the entities are checked against a file containing personal language. If the entities are found in the user-specific language, the two text representations may be fused into a combined text representation and named entity recognition may be performed again to extract further entities.