Dynamic Language Model Adjustment for Voice Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice recognition systems face challenges in balancing accuracy and speed, often producing errors in translating spoken words to text, especially in conversational contexts where n-gram language models fail to account for multiple participants' word usage probabilities, leading to unacceptable mismatches.
Innovation Solution
Implementing a conversation-based language model that adjusts n-gram probabilities by incorporating external text from multiple participants in a text-based conversation, allowing for more accurate voice recognition by reflecting the context and speaker changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a conventional n-gram language model is used for voice recognition, then the system operates at a reasonable speed, but accuracy deteriorates in conversational contexts with multiple participants
Solution Approach 1:
The language model is made dynamic by allowing it to adapt its parameters (n-gram probabilities) in real-time based on the conversational context. The system continuously updates the language model to reflect the current speaker and conversational flow, transforming a static model into a dynamic one that can handle multiple participants effectively
Solution Approach 2:
The invention changes the parameters of the language model by adjusting n-gram probabilities based on external text from conversational partners. This parameter adjustment allows the model to adapt to different speakers and conversational contexts, improving accuracy without requiring a complete model redesign
2Reliability
If the language model is adjusted to account for multiple participants in conversation, then accuracy improves, but computational complexity increases
Solution Approach 1:
The conversational context is segmented by participant, with separate n-gram probability adjustments for each speaker. This segmentation allows the system to track and adapt to individual speaking patterns without requiring a complete overhaul of the language model structure, managing complexity through organized division
3Measurement precision
If more computational resources are spent analyzing sound patterns, then accuracy improves, but processing speed decreases
Solution Approach 1:
The system performs preliminary adjustments to the language model by incorporating external text from conversational partners before voice recognition occurs. This preliminary action prepares the model in advance, reducing the computational burden during actual voice processing and maintaining faster response times
Data Source
AI summary
Embodiments of the present invention provide a method, system and article of manufacture for adjusting a language model within a voice recognition system, based on text received from an external application. The external application may supply text representing the words of one participant to a text-based conversation. In such a case, changes may be made to a language model by analyzing the external text received from the external application.