Neural Speech Masking Signals for Adaptive Privacy Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech masking technologies generate fixed masking signals for different speakers and speech contents, limiting their applicability and user experience in various scenarios.
Innovation Solution
A neural network-based approach is used to dynamically generate masking signals tailored to specific speech contents and masking effects, utilizing a training process that considers factors like speech intelligibility, recognition accuracy, and comfort level, enabling adaptable masking in diverse scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fixed masking signal generation method is used for different speakers and speech contents, then the system complexity is reduced and the implementation is simplified, but the adaptability to different scenarios and speech contents deteriorates, resulting in poor masking effects
Solution Approach 1:
The patent applies dynamics by replacing the fixed masking signal generation method with a dynamic neural network model that adapts to different speakers and speech contents in real-time. The model processes input speech features and generates customized masking signals based on the specific characteristics of each speech instance, enabling the system to adapt to various scenarios while maintaining manageable complexity through automated processing
Solution Approach 2:
The patent changes the parameters of the masking signal generation process by using a neural network model that adjusts masking signal characteristics based on input speech features. Instead of using fixed parameters, the system dynamically modifies masking signal properties (such as frequency content, temporal characteristics, and signal strength) according to the detected speech content, thereby improving adaptability without requiring overly complex manual configuration
2Reliability
If a fixed masking signal generation method is used, then the ease of operation is improved, but the masking effectiveness for different speech contents deteriorates
Solution Approach 1:
The patent applies self-service by enabling the system to automatically analyze speech contents and generate appropriate masking signals without requiring manual intervention or complex user configuration. The neural network model self-adjusts to different speech patterns and automatically optimizes masking effectiveness for each speech instance, maintaining ease of operation while significantly improving masking reliability across different scenarios
Solution Approach 2:
The patent incorporates feedback mechanisms where the system continuously monitors and evaluates the masking effect based on speech content characteristics. The neural network model uses feedback information about the input speech to iteratively refine and adjust the generated masking signals, ensuring optimal masking effectiveness while keeping the user interface simple and easy to operate
Data Source
AI summary
The disclosure relates to the field of communication security and discloses a speech masking method and system, an electronic device, and a non-transitory computer readable storage medium. In the disclosure, after the target speech is obtained, the target speech is not masked according to the traditional fixed masking method, but the target masking effect is determined in advance, and the target masking effect can be determined according to different requirements. After that, the neural network model is trained according to different target masking effects, and the neural network model trained according to the different target masking effects can dynamically provide different masking signals for the target speech. In this way, different masking signals can be generated for different target speeches according to different needs, more scenarios can be applied, and good masking effects can be obtained, thereby improving user experience.