Web ASR Management Tool for Grammar Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech applications require significant effort from experts to map spoken strings of words to specific meanings, with a large sample size of potential utterances needing to be tagged and compared for grammar interpretation tuning, making the recording, analysis, and reporting processes inefficient.
Innovation Solution
The Web ASR Management Tool provides a web interface for capturing, transcribing, and managing utterances, allowing users to create transcription jobs, log calls, and generate reports, including meaning reports for grammar tuning, using Voice eXtensible Markup Language (VXML) and Speech Application Language Tags (SALT) systems, with features like secure access, transcription filtering, and reporting tools.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual mapping of utterances to meanings is performed by experts, then grammar interpretation accuracy is improved, but the time and effort required increases significantly
Solution Approach 1:
The system enables automatic self-tagging of utterances through the ASR engine's transcription capabilities. The engine automatically transcribes spoken input into text and tags it with appropriate metadata, eliminating the need for manual expert tagging while maintaining accuracy through automated analysis and comparison processes.
Solution Approach 2:
The patent replaces the manual mechanical process of expert tagging with an automated computational system. The ASR engine automatically performs transcription, tagging, and comparison operations that previously required human experts to manually map utterances to meanings, significantly reducing time and effort while maintaining or improving accuracy.
2Measurement precision
If a large sample size of utterances is tagged for grammar tuning, then recognition accuracy is improved, but the complexity and resources required increase
Solution Approach 1:
The ASR engine performs multiple functions simultaneously: it transcribes spoken input, tags utterances with metadata, compares transcriptions against grammar rules, and generates tuning recommendations. This multi-functionality consolidates what would otherwise require separate complex systems for recording, transcribing, tagging, and analyzing, reducing overall system complexity while enabling large-scale processing.
Solution Approach 2:
The system enables continuous automated processing of utterances through the ASR engine, which can transcribe and analyze speech data continuously without manual intervention. This continuous automated action allows large sample sizes to be processed efficiently, maintaining recognition accuracy while reducing the operational complexity compared to manual batch processing.
3Loss of information
If manual transcription and reporting processes are used, then detailed analysis is achieved, but productivity decreases
Solution Approach 1:
The ASR engine automatically transcribes spoken input into text and generates structured reports with metadata tagging. This self-service capability eliminates manual transcription while maintaining detailed analysis through automated comparison of transcriptions against grammar rules, significantly improving productivity without sacrificing information quality.
Solution Approach 2:
The patent replaces manual transcription and reporting mechanics with automated computational processes. The ASR engine automatically performs speech-to-text conversion, applies grammar rules, generates structured reports, and provides tuning recommendations, achieving both detailed analysis and high productivity that cannot be simultaneously achieved through manual processes.
Data Source
AI summary
A method, a non-transitory computer readable medium and a system for automatic speech recognition tuning management that comprises, collecting an utterance, analyzing the utterance, correlating the collected utterance to the utterance analysis, and fetching at least one of, the collected utterance, the utterance analysis, and the correlation of the collected utterance to the utterance analysis.