Voice Command Identity Verification via Wake Word Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-user environments, existing voice-controlled devices struggle to accurately distinguish between intended voice commands and unintended speech, leading to potential misinterpretation and inappropriate actions.
Innovation Solution
A voice-controlled device that uses predefined 'wake words' associated with individual users, transitioning between local and network-based speech recognition states to identify and authenticate user commands, thereby differentiating between intended and unintended voice inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the device performs speech recognition on all captured audio signals, then it can respond to any voice command, but it cannot distinguish between intended commands and unintended speech from multiple users
Solution Approach 1:
The system performs preliminary speech recognition to detect wake words before initiating full command processing. The wake word detection acts as a preliminary filter that prepares the system for subsequent user identification and command processing, ensuring that only relevant speech segments undergo complex analysis.
Solution Approach 2:
The system introduces an intermediary user identification step between wake word detection and command execution. This intermediary process analyzes speech characteristics to verify user identity, acting as a mediator that filters out unintended commands while allowing legitimate ones to proceed to execution.
2Adaptability or versatility
If the device responds to all detected voice commands, then it provides comprehensive service, but it executes unintended commands from users who should not be recognized
Solution Approach 1:
The system applies different processing qualities to different speech segments. Wake word detection uses one level of analysis, user identification uses biometric speech characteristic analysis, and command processing uses full speech recognition. Each stage applies the appropriate level of scrutiny based on the local requirements of that processing step.
Solution Approach 2:
The system dynamically adjusts its recognition behavior based on the current interaction context. Once a user is identified through wake word activation, the system maintains that user context for subsequent commands, dynamically switching between user profiles rather than treating all speech uniformly.
3Productivity
If the device processes every speech segment as a potential command, then it ensures no command is missed, but it increases false positive recognition
Solution Approach 1:
The speech processing pipeline is segmented into distinct stages: wake word detection, user identification, and command processing. Each segment handles specific tasks with appropriate precision requirements, preventing the need to apply full command-processing precision to every audio segment and thereby maintaining high productivity while ensuring accuracy where needed.
Solution Approach 2:
The system performs preliminary filtering through wake word detection and user identification before committing to full command execution. This preliminary action discards irrelevant speech segments early in the processing pipeline, allowing high-speed processing of only those segments that pass the preliminary checks, thus maintaining productivity while improving precision.
Data Source
AI summary
Techniques for determining when to perform an action associated with a voice command and when to disregard the voice command. In some instances, the techniques reference an identity of a user that utters a command when making this determination. For instance, if a first user awakens a device or initially begins providing voice commands to the device, the device or another computing device may perform actions associated with subsequent voice commands uttered by the same user, while disregarding voice commands from other users. That is, because the device is engaging in a dialog with a first user, the device may refrain from engaging in a dialog with a different user. Additionally or alternatively, the device may begin a new dialog with the different user, either upon identifying the command of the other user or upon finishing the dialog with the first user.