Voice Recognition Control Unit Using Candidate String Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice recognition systems face challenges in comparing recognition results from multiple units due to differing score values and recognition methods, leading to inadequate normalization and unnecessary processing.
Innovation Solution
A voice recognition apparatus and method that utilize three voice recognition units, where the control unit decides based on recognition results from the first and second units to have the third unit recognize the input voice using candidate character strings from either unit, thereby standardizing scores and preventing unnecessary processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple voice recognition units with different recognition methods and dictionaries are used, then recognition coverage and accuracy are improved, but it becomes impossible to simply compare recognition results due to different score values
Solution Approach 1:
The patent transforms the comparison problem by changing the parameter being compared from score values (which differ between recognition units) to candidate character strings (which are comparable across units). The control unit selects candidate character strings from multiple recognition units and compares these strings directly, bypassing the incomparable score values while maintaining recognition accuracy.
2Ease of operation
If statistical normalization of score values is performed to enable comparison between multiple voice recognition units, then recognition results can be compared, but normalization is insufficient when multiple candidate character strings have different score values
Solution Approach 1:
The patent extracts the candidate character strings from the recognition results of multiple voice recognition units and uses these extracted strings for direct comparison. This extraction approach bypasses the normalization problem entirely, as candidate character strings are inherently comparable regardless of the score values they originated from.
3Measurement precision
If a second stage of voice recognition is always performed to ensure accuracy, then highly accurate results are obtained, but unnecessary processing occurs when the first stage already provides sufficient accuracy
Solution Approach 1:
The patent applies partial action by performing the second stage of voice recognition only when necessary. The control unit determines whether to execute the third voice recognition unit based on the results from the first two units, thereby avoiding unnecessary processing while ensuring accuracy is maintained when needed.
Solution Approach 2:
The system uses feedback from the first two voice recognition units to control whether the third recognition unit is activated. The control unit evaluates the recognition results and score values from the initial units and uses this feedback to decide if additional recognition processing is required, optimizing the balance between accuracy and processing efficiency.
Data Source
AI summary
An object is to provide a technique which can provide a highly valid recognition result while preventing unnecessary processing. A voice recognition device includes first to third voice recognition units, and a control unit. When it is decided based on recognition results obtained by the first and second voice recognition units to cause the third voice recognition unit to recognize an input voice, the control unit causes the third voice recognition unit to recognize the input voice by using a dictionary including a candidate character string obtained by at least one of the first and second voice recognition units.