Voice Template Asset Retrieval for Noise-Resistant Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice-based interface systems face challenges in efficiently retrieving and managing user-specific voice templates, particularly in distinguishing user voices from background noise, which affects the accuracy of voice recognition in diverse work environments.
Innovation Solution
A method that detects an event indicating asset retrieval in a voice-based dialog view, navigates to a built-in asset retrieval workflow activity, retrieves a worker-based voice template, and performs voice template training if none exists, allowing for accurate voice recognition by accessing and storing unique voice templates associated with worker IDs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If voice-based interface systems use generic voice recognition without user-specific templates, then the system complexity is reduced and ease of operation is improved, but voice recognition accuracy deteriorates in diverse work environments with background noise
Solution Approach 1:
The system performs preliminary voice template training during an initialization phase before actual voice recognition tasks. This preliminary action captures user-specific voice characteristics and stores them as templates, enabling accurate voice recognition without requiring complex real-time analysis during operation
Solution Approach 2:
The system creates simplified copies of user voice characteristics in the form of voice templates. These templates are stored and reused for multiple recognition tasks, avoiding the need to reprocess raw voice data each time while maintaining high recognition accuracy
2Measurement precision
If the system implements noise sampling to distinguish user voice from background noise, then voice recognition accuracy is improved, but the time required for setup and calibration increases
Solution Approach 1:
The system performs noise sampling and voice characteristic capture during an initial setup phase, storing the results as voice templates. This preliminary action eliminates the need for repeated calibration during normal operation, reducing ongoing setup time while maintaining accurate voice distinction
Solution Approach 2:
The voice template training process automatically captures and processes voice samples without requiring manual intervention or complex user configuration. The system self-adjusts to the user's voice characteristics, minimizing the time and effort required for setup
3Measurement precision
If the system retrieves and stores unique voice templates for each worker ID, then voice recognition accuracy is improved and noise interference is reduced, but the complexity of asset management and template retrieval increases
Solution Approach 1:
The system introduces an asset management intermediary layer that handles voice template storage, retrieval, and management. This intermediary abstracts the complexity of managing multiple voice templates behind a simple interface that uses worker IDs as keys, enabling accurate voice recognition without burdening the user with asset management complexity
Data Source
AI summary
A method includes detecting an event published to a workflow activity by a voice based dialog view, wherein the event indicates a state of asset retrieval, navigating to a built-in asset retrieval work activity, retrieving an asset, and dismissing the workflow activity to revert to a workflow activity associated with the voice based dialog view.