Smartphone Sensor Voice Activity Detection for Cognitive Change
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting cognitive changes are hindered by privacy and technical obstacles related to collecting voice data, which limits the ability to effectively monitor cognitive capabilities.
Innovation Solution
A system utilizing sensors such as accelerometers, gyroscopes, and magnetometers within a mobile phone to detect voice activity and determine cognitive changes without directly recording the user's voice, using machine learning algorithms to analyze sensor signals and generate voice activity datasets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If voice data is collected to detect cognitive changes, then measurement precision is improved, but privacy concerns and technical obstacles worsen
Solution Approach 1:
The patent uses smartphone sensors (accelerometer, gyroscope, magnetometer) as intermediaries to detect voice activity indirectly through vibrations caused by speech, rather than directly recording voice data. This mediator approach enables cognitive change detection while avoiding direct voice collection, thus resolving the privacy concern.
Solution Approach 2:
The patent replaces the acoustic measurement system (microphone recording voice) with a mechanical vibration detection system (accelerometer, gyroscope, magnetometer detecting phone vibrations from speech). This substitution maintains measurement capability while eliminating privacy issues associated with voice recording.
2Measurement precision
If voice data is collected to detect cognitive changes, then measurement precision is improved, but technical obstacles worsen
Solution Approach 1:
The patent uses smartphone sensors (accelerometer, gyroscope, magnetometer) as intermediaries to detect voice activity indirectly through vibrations caused by speech, rather than directly recording voice data. This mediator approach enables cognitive change detection while avoiding direct voice collection, thus resolving the privacy concern.
Solution Approach 2:
The patent replaces the acoustic measurement system (microphone recording voice) with a mechanical vibration detection system (accelerometer, gyroscope, magnetometer detecting phone vibrations from speech). This substitution maintains measurement capability while eliminating privacy issues associated with voice recording.
3Measurement precision
If sensor signals are processed using machine learning algorithms, then cognitive capability determination accuracy is improved, but device complexity increases
Solution Approach 1:
The patent extracts specific relevant features from sensor signals (voice activity detection, signal segments) rather than processing entire raw signal datasets. This partial action approach achieves sufficient cognitive capability determination accuracy while reducing computational complexity.
Solution Approach 2:
The patent divides sensor signals into discrete segments and processes them individually through machine learning algorithms. This segmentation enables manageable computation while maintaining detection accuracy through systematic analysis of signal portions.
Data Source
AI summary
Generally, systems and methods for determining a change of a cognitive capability of a user are disclosed. The method may include: receiving at least one sensor signal acquired by at least one sensor (such as an accelerometer, gyro and/or magnetometer) mounted within a mobile phone of the user; determining a voice activity dataset based on the at least one sensor signal; and determining a change of a cognitive capability of the user based on the voice activity dataset. Advantageously, the disclosed systems and methods may enable determining anomalies and trends in the cognition of the user based on the sensor(s) mounted within the mobile phone of the user, without collecting and/or recording the voice of the user.