Automatic Speech Recognition Grammar Tuning via Automated Difference Reporting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Automatic Speech Recognition (ASR) systems require significant effort from subject matter experts to map spoken strings of words to specific meanings, necessitating a comparison between transcribed files and semantic interpretation grammar for tuning, which is inefficient due to the large sample size of potential utterances.
Innovation Solution
A computer system and method that compares transcribed utterances to semantic interpretation grammar, using a graphical user interface to facilitate reporting of differences, allowing for the tuning of the grammar by integrating with a transcription system and providing a flexible software solution for analyzing and reporting these differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual comparison of transcribed utterances to semantic interpretation grammar is performed by subject matter experts, then tuning accuracy can be achieved, but the process requires significant time and manual effort due to the large sample size of potential utterances
Solution Approach 1:
The system enables automatic self-tuning by comparing transcribed utterances against the semantic interpretation grammar without requiring manual review by subject matter experts. The computer automatically identifies mismatches and suggests corrections, allowing the ASR system to improve itself autonomously.
Solution Approach 2:
The patent replaces the manual mechanical process of expert review with an automated computer-based system that performs the comparison and analysis. This substitution eliminates the need for human experts to manually examine each utterance while maintaining or improving the tuning quality.
2Reliability
If the sample size of potential utterances for tagging is increased to improve grammar accuracy, then the quality of semantic interpretation improves, but the complexity and effort required for the Reporting process increases
Solution Approach 1:
The automated system handles the complexity of processing large utterance samples through self-service mechanisms, where the computer automatically performs the comparison, analysis, and reporting without requiring proportional increases in human expert resources. The system manages its own complexity internally while presenting simplified outputs.
3Measurement precision
If manual tuning of semantic interpretation grammar is performed to improve ASR accuracy, then recognition quality improves, but productivity decreases due to the extensive manual review required
Solution Approach 1:
The system performs automatic self-tuning by comparing transcribed utterances against the semantic interpretation grammar and generating improvement suggestions without requiring manual intervention for each utterance. This self-service capability dramatically increases tuning throughput while maintaining quality improvements.
Solution Approach 2:
The patent replaces the manual tuning process with an automated computer-based system that can process and analyze large numbers of utterances rapidly. This substitution maintains the quality improvements of manual tuning while increasing productivity by eliminating the time constraints of human review.
Data Source
AI summary
A method, a computer readable medium and a system for reporting automatic speech recognition that comprises, collecting an utterance, analyzing the utterance, receiving a translation of the utterance, and determining a difference between the analyzed utterance and the translated utterance. An embodiment the disclosure includes updating the utterance analysis based upon the determined difference, correlating the analyzed utterance to the translated utterance and tracking determined difference by a translator. In another embodiment the disclose includes reporting, categorizing, sorting, and grouping the determined difference.