Voice Trigger Avoidance Using Thermal Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Voice-controlled devices face challenges in distinguishing between human-generated voice commands and similar sounds from audio sources like televisions or computers, leading to potential misinterpretation and unintended activation.
Innovation Solution
The system employs thermal signature analysis using infrared sensors and machine learning models to determine if a human is present near the device, thereby verifying the authenticity of voice commands by emitting and detecting thermal signals to differentiate between human and non-human sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If voice-controlled devices use audio data alone to trigger commands, then the device responds quickly to voice commands, but it cannot distinguish between human speech and sounds from audio devices
Solution Approach 1:
The patent combines audio data processing with thermal signature analysis to create a hybrid verification system. The voice-controlled device integrates multiple sensing modalities (acoustic and thermal) to simultaneously detect and verify voice commands, resolving the contradiction by merging the speed of audio processing with the accuracy of thermal verification to distinguish human speech from audio device sounds.
Solution Approach 2:
The patent introduces thermal signature analysis as an intermediary verification mechanism between audio detection and command execution. This intermediary layer processes thermal data to confirm the presence of a human user before executing voice commands, thereby improving reliability without requiring complete redesign of the original audio-processing system.
2Reliability
If the device uses thermal signature analysis to verify human presence, then false activations are reduced, but the system complexity increases
Solution Approach 1:
The patent applies thermal signature analysis selectively rather than continuously - only when audio data is detected that might require verification. This partial application of the thermal verification mechanism reduces the overall system complexity burden while maintaining high reliability for distinguishing human speech from audio device sounds.
Solution Approach 2:
The thermal signature analysis serves as an intermediary verification layer that processes thermal data independently from audio processing. This modular intermediary approach allows the system to maintain separate processing streams for audio and thermal data, reducing the complexity burden by avoiding complete system redesign while still achieving improved verification accuracy.
3Productivity
If the device processes audio data without thermal verification, then the response time is fast, but unintended activation from audio devices occurs
Solution Approach 1:
The patent performs preliminary thermal signature detection and analysis before finalizing voice command processing. By conducting thermal verification in advance or concurrently with audio processing, the system maintains fast response times while preventing unintended activation from audio devices through pre-verification of human presence.
Solution Approach 2:
The patent merges audio processing and thermal verification into a concurrent processing architecture where both data streams are handled simultaneously. This merging approach allows the system to maintain fast response speeds by processing both modalities in parallel rather than sequentially, while still achieving reliable distinction between human speech and audio device sounds through combined analysis.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively filters out non-human audio inputs, ensuring that only valid voice commands from nearby humans are processed, reducing false activations and enhancing the accuracy of voice-controlled systems.
Implementation Method 1
The voice controlled device may determine whether the audio data is indicative of user input from a user nearby by using a thermal signature to determine whether a person is located near the voice controlled device
Data Source
AI summary
Methods and systems for processing voice commands are disclosed. A voice controlled device may receive audio data comprising a voice command. Location information indicative of the source of the audio data may be determined. One or more devices may be caused to determine signals based on the location information. The one or more devices may receive thermal data in response to the signals. The thermal data may be analyzed to determine if the thermal data indicates the presence of a person at the expected location. If a person is detected, then the audio data may processed to cause the voice command to be executed.