HMD Motion Sensor Filtering for Voice Privacy Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Motion sensors on head-mounted displays can inadvertently capture speech vibrations, potentially violating user privacy by allowing eavesdropping even when microphones are off.
Innovation Solution
Implement a system with motion sensors and processors that receive tuning signals to remove or obfuscate voice-caused components from sensor signals, using techniques like signal subtraction, bandpass filtering, and noise addition, optionally enhanced by machine learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion sensors are used to track head movements for VR/AR applications, then motion tracking accuracy is improved, but user privacy is compromised due to ability to capture speech vibrations
Solution Approach 1:
The patent extracts and removes voice-related components from motion sensor signals through signal processing. The system identifies frequency ranges associated with speech vibrations and selectively removes or obfuscates these components while preserving motion tracking data, thereby eliminating the privacy hazard while maintaining measurement precision.
Solution Approach 2:
The patent introduces an intermediary signal processing layer between the motion sensor and the application. This intermediary processes the raw sensor signals to remove voice components before passing cleaned signals to the application, acting as a mediator that protects privacy while preserving necessary motion data.
2Object-affected harmful factors
If voice components are completely removed from motion sensor signals, then privacy protection is improved, but motion sensor functionality is degraded
Solution Approach 1:
The patent applies local quality by selectively processing only specific frequency ranges associated with voice components. Instead of removing all signal components, the system targets specific frequency bands where speech vibrations occur, removing or obfuscating only those local portions while leaving the rest of the motion signal intact and functional.
Solution Approach 2:
The patent implements partial action by offering adjustable levels of voice component removal. The system can remove voice components to varying degrees based on privacy requirements, allowing users to balance privacy protection with functionality needs. The processing is applied partially to only the necessary frequency ranges rather than the entire signal spectrum.
Data Source
AI summary
To protect a user's privacy by reducing a malicious developer's ability to eavesdrop on unwitting HMD users by converting signals from a motion sensor in the HMD to speech or speaker recognition, a microphone can record ambient sound and voice which is subtracted from the motion sensor data before the sensor data is made available to the game developer. In another technique, a band pass filter subtracts frequency in the sensor signals within the voice range. Still a third technique blends statistical noise into the motion sensor signal before passing to game developers to obfuscate the user's speech. The amount by which voice components in the motion signal are eliminated or obfuscated can be tuned by a person or app.