Speech Recognition Pre-Processing Apparatus for Trailing Silence Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional speech recognition devices face degraded performance due to inconsistent trailing silence periods, which are influenced by user tendencies, leading to inefficient resource consumption and inaccurate speech recognition in portable devices like navigation systems.
Innovation Solution
A pre-processing apparatus and method that detects and adjusts the trailing silence period in speech signals based on a reference period, using a detection unit, storage unit, and adjusting unit to ensure optimal alignment with user input after a start sound, thereby improving speech recognition functionality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a trailing silence period is applied to detect speech input period, then speech input detection is improved, but speech recognition rate is degraded due to user tendency variations
Solution Approach 1:
The patent applies dynamics by making the trailing silence period adjustable rather than fixed. The system dynamically adapts the trailing silence period based on detected user tendencies, allowing the parameter to change according to individual user patterns while maintaining accurate speech input detection.
Solution Approach 2:
The patent implements parameter changes by modifying the trailing silence period duration based on detected user tendencies. The system analyzes multiple speech inputs to identify user-specific patterns and adjusts the trailing silence period parameter accordingly, improving recognition accuracy for each user.
2Reliability
If speech recognition processing is performed, then speech understanding is improved, but resource consumption increases in portable devices
Solution Approach 1:
The patent applies preliminary action by performing pre-processing on speech signals before full recognition processing. The system detects user tendencies in advance and adjusts parameters like trailing silence period beforehand, so that the main recognition process can operate more efficiently with pre-optimized parameters.
Solution Approach 2:
The patent implements segmentation by dividing speech recognition processing into distinct stages: pre-processing (detecting user tendencies, adjusting parameters), main recognition processing, and post-processing. This segmentation allows resource-intensive operations to be minimized in the main processing stage while maintaining high recognition accuracy.
3Quantity of substance
If trailing silence period is increased to capture complete speech, then speech input coverage is improved, but speech recognition accuracy is degraded
Solution Approach 1:
The patent applies parameter changes by adjusting the trailing silence period duration based on detected user tendencies. Instead of using a fixed or uniformly increased trailing silence period, the system dynamically optimizes this parameter for each user, capturing sufficient speech signal while maintaining recognition accuracy.
Data Source
AI summary
A pre-processing apparatus for speech recognition may include: a trailing silence period detection unit configured to detect the length of a trailing silence period contained in a speech signal; a reference trailing silence period storage unit configured to store the length of a reference trailing silence period; and a trailing silence period adjusting unit configured to adjust the length of the trailing silence period contained in the speech signal based on the length of the reference trailing silence period.