Voice Identification Confidence Adjustment for Noise Scenarios
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice identification technologies, such as voice assistant software, perform poorly in noisy environments, leading to reduced identification rates and user experience issues, as they fail to effectively adjust confidence values to account for noise scenarios.
Innovation Solution
A method and apparatus that obtain a noise scenario and adjust confidence values based on pre-stored empirical data, allowing for flexible adjustment of confidence thresholds to improve voice identification accuracy in noisy environments by determining the noise type and magnitude through frequency cepstrum coefficients and Gaussian mixture models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a fixed confidence threshold is used for voice identification, then the identification process is simple and fast, but the voice identification rate deteriorates in noisy environments
Solution Approach 1:
The patent implements dynamic confidence threshold adjustment by introducing a noise scenario recognition module that automatically adapts the confidence threshold based on the detected noise level. The system transitions from a static threshold to a dynamic one that changes according to environmental noise conditions, thereby maintaining high identification rates across varying acoustic environments without requiring manual intervention
Solution Approach 2:
The patent changes the parameter of confidence threshold from a fixed value to a variable that depends on noise scenario parameters. By introducing noise level detection and mapping noise levels to corresponding confidence threshold adjustments, the system optimizes identification accuracy for different acoustic conditions while managing complexity through predefined threshold tables
2Object-affected harmful factors
If noise reduction processing is applied to voice data, then some noise interference is reduced, but the identification rate may be lowered due to processing distortions
Solution Approach 1:
The patent applies noise reduction processing as a preliminary step before voice identification, but crucially adjusts the confidence threshold afterward to compensate for any distortions introduced. This two-stage approach - first reducing noise, then adjusting the decision criterion - allows the system to benefit from noise reduction while compensating for its potential negative effects on identification accuracy
Solution Approach 2:
The patent introduces a feedback mechanism where the noise scenario recognition results are used to adjust the confidence threshold in response to detected noise conditions. This closed-loop approach allows the system to adapt to actual noise levels and processing effects, maintaining reliability by compensating for distortions through dynamic threshold adjustment based on real-time noise assessment
Data Source
AI summary
Embodiments of the present invention provide a voice identification method, including: obtaining voice data; obtaining a first confidence value according to the voice data; obtaining a noise scenario according to the voice data; obtaining a second confidence value corresponding to the noise scenario according to the first confidence value; and if the second confidence value is greater than or equal to a pre-stored confidence threshold, processing the voice data. An apparatus is also provided. The method and apparatus that flexibly adjust the confidence value according to the noise scenario greatly improve a voice identification rate under a noise environment.