Wakeword Detection Segmentation for Battery Conservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech recognition systems face challenges in accurately detecting wakewords and efficiently managing battery usage, as they often rely on a single wakeword detector that requires extensive training and continuous activation, leading to increased computational resources and battery drain.
Innovation Solution
Implementing multiple wakeword detectors on a single device, where the initial detector is trained by the user and later replaced by a companion application's detector, which is trained using user speech and deactivates the user-trained model to conserve battery life by reducing computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single wakeword detector is continuously activated to ensure reliable detection, then detection reliability is improved, but battery consumption increases
Solution Approach 1:
The system segments the wakeword detection function into two distinct detectors: a first wakeword detector that remains continuously active with low computational requirements, and a second wakeword detector that is activated only when the first detector identifies a potential wakeword. This segmentation allows the system to maintain reliable detection while minimizing battery consumption by keeping the more resource-intensive second detector dormant most of the time.
2Measurement precision
If a user-trained wakeword detector model is continuously active to provide accurate detection, then detection accuracy is improved, but computational resources are overconsumed
Solution Approach 1:
The system applies partial action by having the first wakeword detector perform a preliminary, less computationally intensive screening of audio input. Only when this partial detection suggests a potential wakeword is present does the system activate the second, more accurate but resource-intensive detector. This approach achieves high detection accuracy when needed while avoiding continuous overconsumption of computational resources.
Data Source
AI summary
Techniques for implementing multiple wakeword detectors on a single device are described. A digital signal processor (DSP) of the device may initially include an untrained wakeword detection component. The wakeword detection component of the DSP may be trained by engaging a user to speak particular utterances. Once a companion application is configured to implement a wakeword detection component, the companion application's wakeword detection component may be trained specific to the user of the device. Once the companion application's wakeword detection component is trained, the DSP wakeword detection component may be deactivated or its accuracy adjusted.