Speech Recognition Error Detection via Alternative Result Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automatic speech recognition (ASR) systems often produce recognition errors due to low-quality audio, unclear speech, or errors in analysis, which can lead to serious consequences, especially in domains like medicine, where misinterpretations can result in incorrect diagnoses or treatments.
Innovation Solution
A method is introduced to evaluate recognition results by comparing the top recognition result to alternative results, using sets of confusable words or phrases, and employing language models to identify potential significant errors, triggering alerts for reviewers to correct errors that could have significant consequences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If ASR systems process speech input to generate recognition results, then speech-to-text conversion is achieved, but recognition errors occur due to low-quality audio, unclear speech, or analysis errors
Solution Approach 1:
The patent introduces an evaluation engine as an intermediary component that receives recognition results from the ASR system and compares multiple alternative results. This mediator analyzes discrepancies between results and identifies potential errors before final output, thereby improving reliability without sacrificing the productivity of the original ASR system
Solution Approach 2:
The system implements a feedback mechanism where recognition results are evaluated by comparing alternative interpretations. The evaluation engine provides feedback by identifying discrepancies and potential errors, allowing the system to self-correct or flag uncertain results for manual review, thus improving recognition accuracy while maintaining efficient processing
2Reliability
If multiple alternative recognition results are generated and compared, then potential errors can be identified, but system complexity increases
Solution Approach 1:
The evaluation engine applies partial action by not evaluating all possible recognition results exhaustively. Instead, it selectively compares alternative results and focuses on identifying specific types of discrepancies (such as semantically significant differences), achieving reliable error detection without the computational overhead of complete analysis
Solution Approach 2:
The patent extracts the error detection function from the main ASR processing pipeline by implementing a separate evaluation engine. This extraction allows the core ASR system to remain simple and efficient while the evaluation component handles the complex task of comparing alternative results and identifying potential errors
Data Source
AI summary
In some embodiments, the recognition results produced by a speech processing system (which may include two or more recognition results, including a top recognition result and one or more alternative recognition results) based on an analysis of a speech input, are evaluated for indications of potential significant errors. In some embodiments, the recognition results may be evaluated using one or more sets of words and/or phrases, such as pairs of words/phrases that may include words/phrases that are acoustically similar to one another and/or that, when included in a result, would change a meaning of the result in a manner that would be significant for a domain. The recognition results may be evaluated using the set(s) of words/phrases to determine, when the top result includes a word/phrase from a set of words/phrases, whether any of the alternative recognition results includes any of the other, corresponding words/phrases from the set.