Natural Language Processing Accuracy via Speech Engine Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech recognition technologies suffer from low accuracy in natural language processing, as a single engine is responsible for both automatic speech recognition and semantic understanding, often failing to recognize or identify errors effectively.
Innovation Solution
A system utilizing multiple speech recognition engines connected to a natural language input device, with a data fusion model to identify the correct result from multiple outputs and a separate semantic understanding model for processing, trained using a corpus of results to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single speech recognition engine is used for both automatic speech recognition and semantic understanding, then device complexity is reduced, but natural language processing accuracy deteriorates
Solution Approach 1:
The system divides the speech processing function into separate modules: multiple speech recognition engines handle automatic speech recognition independently, while a separate semantic understanding model processes the recognized speech. This segmentation allows each component to specialize in its specific task, improving overall accuracy while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The system combines multiple speech recognition engines to process the same input simultaneously, merging their outputs through a data fusion model. This combination allows the system to leverage diverse recognition capabilities and select the most accurate result, thereby improving natural language processing accuracy without requiring a single overly complex engine.
2Measurement precision
If multiple speech recognition engines are used, then natural language processing accuracy is improved, but device complexity increases
Solution Approach 1:
A data fusion model acts as an intermediary between multiple speech recognition engines and the semantic understanding model. This mediator receives outputs from multiple engines, applies fusion algorithms to determine the most accurate result, and passes the refined output forward. The intermediary manages the complexity of coordinating multiple engines while enabling accuracy improvements.
Solution Approach 2:
The system implements feedback mechanisms where the data fusion model evaluates results from multiple speech recognition engines and selects or combines the most accurate outputs. This feedback loop allows the system to learn from multiple recognition approaches and continuously improve accuracy by leveraging the strengths of different engines without requiring proportional increases in overall system complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system and method for improving accuracy of natural language processing using a plurality of speech recognition engines, a data fusion model to identify a correct result from the plurality of speech recognition engines and a semantic understanding model, separate and distinct from the speech recognition model, to process the correct results. A corpus is developed using the correct results and the corpus is used to train the data fusion model and the semantic understanding model.