Accelerometer-Based Hotword Detection for Voice Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice-control systems consume significant power due to continuous microphone usage for hotword detection, which is inefficient for battery-powered devices like smartphones and smartwatches.
Innovation Solution
The AccelWord system employs a MEMS accelerometer to detect hotwords by analyzing accelerometer data, using low-complexity features and a machine-learning classifier to activate the voice-control system only when a hotword is detected, thereby reducing power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If continuous hotword detection is performed using a microphone, then hotword detection accuracy is maintained, but power consumption increases significantly
Solution Approach 1:
The patent introduces an accelerometer as an intermediary sensor to perform preliminary hotword detection. The accelerometer detects acoustic vibrations through the device structure, providing a low-power screening mechanism that triggers the high-power microphone only when a potential hotword is detected, thus maintaining detection accuracy while reducing overall power consumption
Solution Approach 2:
The system performs preliminary hotword detection using the accelerometer before activating the microphone. By pre-screening for hotwords through the accelerometer's vibration sensing capability, the system prepares in advance and only engages the power-intensive microphone when necessary, resolving the contradiction between continuous detection accuracy and power consumption
2Speed
If the microphone is continuously powered on for hotword detection, then detection responsiveness is improved, but energy efficiency deteriorates
Solution Approach 1:
The system implements periodic action by having the accelerometer continuously monitor for hotwords in a low-power state, and only activating the microphone periodically when the accelerometer detects a potential hotword event. This periodic activation pattern maintains detection responsiveness while dramatically improving energy efficiency compared to continuous microphone operation
3Use of energy by moving object
If accelerometer-based hotword detection is implemented, then power consumption is reduced, but detection accuracy may be affected by user mobility and noise
Solution Approach 1:
The patent merges the detection capabilities of two different sensors: the accelerometer for initial hotword screening and the microphone for confirmation and subsequent voice processing. This combination allows the system to leverage the low-power advantage of accelerometer-based detection while using the microphone to verify detections and handle cases where accelerometer accuracy is compromised by mobility or noise
Solution Approach 2:
The system replaces the traditional acoustic field-based detection (microphone) with a mechanical vibration-based detection (accelerometer) for the initial hotword screening stage. This substitution enables low-power operation while the mechanical vibration sensing complements rather than completely replaces the acoustic sensing, maintaining accuracy across different conditions
Data Source
AI summary
The disclosed embodiments relate to the design of a system that uses an accelerometer in a mobile device to detect hotwords, which activate a voice interface to recognize subsequent voice input. During operation, the system gathers samples comprising readings from the accelerometer in the mobile device. Next, the system calculates features from the gathered samples. The system then classifies the calculated features using a classifier, which has been trained to detect hotwords based on calculated features. Finally, if the classifier detects one or more hotwords, the system causes the mobile device to launch a voice-control system that recognizes subsequent voice input received from an audio microphone.