Speech Recognition Activation via Voice and Command Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Always-on automatic speech recognition (ASR) systems in mobile devices consume high power, reducing battery life, and existing solutions like reduced-capacity keyword detection suffer from high false activation rates due to misidentification of background noise or ordinary speech as user commands.
Innovation Solution
Implementing a pre-stage with a user voice verifier and a speaking target detector to selectively activate the ASR engine only when an input sound is identified as a registered user's voice and a command or question, using less complex voice and tone models to reduce power consumption and false activations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If an always-on ASR system is used to continually scan audio input, then speech recognition capability is improved, but power consumption increases
Solution Approach 1:
The speech recognition system is divided into two distinct stages: a reduced-capacity keyword detection stage that operates continuously with low power consumption, and a full-capacity ASR engine that activates only when keywords are detected. This segmentation allows the system to maintain speech recognition capability while significantly reducing overall power consumption by keeping the high-power ASR engine dormant most of the time.
Solution Approach 2:
The keyword detection stage performs preliminary screening of audio input continuously before activating the full ASR engine. By pre-processing the audio signal and detecting keywords in advance, the system can selectively activate the power-intensive ASR engine only when necessary, thereby reducing overall power consumption while maintaining responsiveness.
2Use of energy by moving object
If reduced-capacity keyword detection is used to reduce power consumption, then power consumption is reduced, but false activation rate increases
Solution Approach 1:
The keyword detection stage serves as an intermediary between the audio input and the full ASR engine. This intermediate layer filters audio signals by detecting specific keywords, allowing the system to activate the full ASR engine only when relevant keywords are present. This intermediary mechanism reduces false activations by preventing the full ASR engine from processing unrelated audio content while maintaining low power consumption.
3Reliability
If the ASR engine is activated frequently to ensure no commands are missed, then speech recognition reliability is improved, but processing resource consumption increases
Solution Approach 1:
Instead of continuously activating the full ASR engine, the system applies partial action by using a reduced-capacity keyword detection stage that operates continuously with minimal resources. The full ASR engine is activated excessively only when keywords are detected, which is sufficient to catch all relevant commands while avoiding unnecessary processing of unrelated audio, thus optimizing processing resource efficiency.
Data Source
AI summary
A device to process an audio signal representing input sound includes a user voice verifier configured to generate a first indication based on whether the audio signal represents a user's voice. The device includes a speaking target detector configured to generate a second indication based on whether the audio signal represents at least one of a command or a question. The device includes an activation signal unit configured to selectively generate an activation signal based on the first indication and the second indication. The device also includes an automatic speech recognition engine configured to be activated, responsive to the activation signal, to process the audio signal.