Accelerometer-Assisted Hot Word Detection for Wearables

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Electronic devices face high power consumption and false activation issues due to the difficulty in accurately detecting specific user voices in noisy environments, leading to unnecessary power drain and incorrect triggering of hot word detection features.

Innovation Solution

Incorporating an accelerometer to differentiate user voice input from background noise by comparing accelerometer signals to a predetermined threshold, which activates the microphone only when the user is speaking, thereby reducing false detections and conserving battery life.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the device continuously monitors voice signals for hot word detection, then the detection capability is maintained, but power consumption increases

Engineering Contradiction:

Improvehot word detection capabilityVSAvoidpower consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system transitions from continuous monitoring to periodic monitoring by using the accelerometer to detect user presence and activity states. The microphone is activated only during periods when the accelerometer indicates the user is present and potentially speaking, creating an on-demand monitoring scheme that reduces overall power consumption while maintaining detection capability when needed.

Inventive Principle:

Principle #19Periodic action

2Measurement precision

If the device activates the microphone continuously to detect user voice, then voice detection accuracy is improved, but battery life is reduced

Engineering Contradiction:

Improvevoice detection accuracyVSAvoidbattery life

Core Design Contradiction:

Measurement precisionVSDuration of action of moving object

Solution Approach 1:

The accelerometer continuously monitors for user presence and activity in advance before voice detection is needed. By detecting user presence and potential speaking states ahead of time, the system can pre-activate the microphone only when necessary, ensuring accurate voice detection when the user is actually present while avoiding unnecessary microphone activation that would drain the battery.

Inventive Principle:

Principle #10Preliminary action

3Device complexity

If the device uses only microphone signals for hot word detection, then the system is simple, but false detection by other people's voices increases

Engineering Contradiction:

Improvedetection system complexityVSAvoidfalse detection rate

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The system merges accelerometer data with microphone signals to create a multi-modal detection approach. The accelerometer provides contextual information about user presence and activity that complements the audio signals. By combining these two data sources, the system can distinguish between the user's voice and other people's voices more reliably, reducing false detections while maintaining system simplicity through the use of existing sensor components.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS11232800B2Personalized talking detector for electronic device

Publication Date: 2022.01.25 GOOGLE LLC

AI summary

The present disclosure provides for improved hot word detection in electronic devices, particularly small form factor devices such as wearables. The electronic device includes an accelerometer onboard to pick up voice in noisy conditions, and utilizes the accelerometer to confirm that a particular user intended to activate the hot word detection, thereby reducing false detection of other people's voices.