Wake Word Self-Triggering Avoidance via Dual Recognizer Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice command-and-control systems often mistakenly recognize audio output from devices as user input, leading to self-triggering and a frustrating user experience due to reflections of audio waves.
Innovation Solution
Implementing two separate wake word recognizers, one for microphone input and another for audio output, with a reconciliation module to determine if the wake word detected by the microphone is actually from the device itself, allowing the system to ignore self-triggering instances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the device uses wake word recognition to enable hands-free voice control, then user convenience is improved, but the system becomes prone to self-triggering errors
Solution Approach 1:
The wake word recognition system is segmented into two independent recognizers: a first recognizer that processes microphone input and a second recognizer that processes audio-out signal. This segmentation allows the system to separately track wake word detections from user input versus device output, enabling the reconciliation module to distinguish between valid user commands and self-triggered false positives.
Solution Approach 2:
The system implements feedback by having the second recognizer monitor the audio-out signal for wake word content, then feeding this information back to the reconciliation module. This feedback loop enables the system to recognize when the device itself is outputting the wake word and prevent erroneous self-triggering while maintaining normal voice control functionality.
2Speed
If the device continuously monitors for wake word to respond to user commands, then responsiveness is improved, but false detection of self-output audio occurs
Solution Approach 1:
The detection system is segmented into two parallel recognition paths: one for microphone input and one for audio-out signal. Both paths operate simultaneously to maintain responsiveness, but the segmentation allows precise differentiation between user-spoken wake words and device-generated audio, eliminating false detections while preserving fast response capability.
Solution Approach 2:
The reconciliation module serves as an intermediary that receives wake word detection results from both recognizers and mediates between them. It uses the audio-out recognizer's detection of wake word in the output signal as a marker to identify and filter out false positives from the microphone input recognizer, thereby improving detection precision without sacrificing speed.
3Productivity
If the system processes all wake word detections to maintain responsiveness, then user experience is improved, but unnecessary processing occurs due to self-triggering
Solution Approach 1:
The second recognizer performs preliminary action by analyzing the audio-out signal for wake word content before the full processing pipeline is activated. This preliminary detection allows the system to identify self-triggered wake words in advance and prevent unnecessary downstream processing, thereby reducing energy consumption while maintaining productivity through efficient early filtering.
Solution Approach 2:
The system uses feedback from the second recognizer to modulate the processing pipeline. When the audio-out recognizer detects wake word in the output signal, it provides feedback to the reconciliation module to suppress or ignore corresponding wake word detections from the microphone, preventing wasteful processing of self-triggered commands while preserving full processing throughput for valid user commands.
Data Source
AI summary
Techniques for avoiding wake word self-triggering are provided. In one embodiment, an electronic device can receive an audio-out signal to be output as audio via a speaker of the device and can attempt to recognize a wake word in the audio-out signal using a first recognizer. If the wake word is recognized in the audio-out signal, the electronic device can further determine whether a wake word match is made using a second recognizer with respect to a mic-in audio signal captured via a microphone of the device at approximately the same time that the audio-out signal is output via the speaker. If so, the electronic device can ignore the wake word match made using the second recognizer.