Speech Recognition Power Management via Keyword Triggering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech recognition systems in computing devices often maintain a persistently active state, leading to excessive energy consumption, especially in mobile devices, as they continuously connect to networks and power hardware for speech recognition capabilities, which is inefficient and problematic for battery-powered devices.
Innovation Solution
A power management subsystem that selectively activates and deactivates modules within a computing device based on audio input, using keywords (wakewords and sleepwords) to manage power consumption, including a network interface module, speech detection, and processing units, allowing for efficient energy use by activating components only when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If speech recognition capabilities are maintained in a persistently active state, then speech recognition responsiveness is improved, but energy consumption increases
Solution Approach 1:
The system dynamically transitions speech recognition capabilities between active and inactive states based on operational conditions. The speech recognition module is activated when responsiveness is needed and deactivated when energy conservation is prioritized, making the system adaptable rather than static in its resource allocation.
Solution Approach 2:
The system employs periodic activation of speech recognition capabilities rather than continuous operation. By cycling between active and inactive states based on triggers or time intervals, the system maintains functionality when needed while reducing overall energy consumption during periods when speech recognition is not required.
2Measurement precision
If network interface module remains connected, then speech recognition accuracy is improved, but power consumption increases
Solution Approach 1:
The network interface module dynamically adjusts its connectivity state based on operational needs. The system connects to the network when enhanced speech recognition accuracy through server processing is required and disconnects when basic local processing suffices, optimizing the balance between accuracy and power consumption.
Solution Approach 2:
The system performs preliminary assessment to determine whether network connectivity is needed before establishing the connection. By evaluating whether local processing capabilities are sufficient or if server-based processing is required, the system avoids unnecessary network connections and associated power consumption.
3Reliability
If hardware for speech recognition is continuously powered, then recognition capability is maintained, but energy efficiency deteriorates
Solution Approach 1:
The speech recognition hardware dynamically transitions between powered and unpowered states based on system conditions. When speech recognition functionality is required, the hardware is activated; when it is not needed, the hardware is deactivated to conserve energy, maintaining reliability only when necessary.
Solution Approach 2:
The system employs periodic activation of speech recognition hardware rather than continuous powering. By cycling the hardware between active and inactive states based on operational requirements, the system maintains recognition capability when needed while significantly improving energy efficiency during inactive periods.
Data Source
AI summary
Power consumption for a computing device may be managed by one or more keywords. For example, if an audio input obtained by the computing device includes a keyword, a network interface module and/or an application processing module of the computing device may be activated. The audio input may then be transmitted via the network interface module to a remote computing device, such as a speech recognition server. Alternately, the computing device may be provided with a speech recognition engine configured to process the audio input for on-device speech recognition.