Input Detection Windowing for Voice Assistant False Positives

Solution Overview

Problem

Traditional wake-word engines in network microphone devices (NMDs) are prone to false positives due to misidentification of wake words in ambient audio, leading to unnecessary resource consumption and potential interruptions in media playback.

Innovation Solution

Implementing a keyword engine that listens for specific keywords during a designated input detection window, triggered by events such as changes in media content or user interaction, reduces false positives by requiring both keyword detection and specific playback conditions before generating a command.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If wake-word engines continuously monitor ambient audio for wake words, then voice command responsiveness is improved, but false positive rate increases

Engineering Contradiction:

Improvevoice command responsivenessVSAvoidfalse positive rate

Core Design Contradiction:

SpeedVSReliability

Solution Approach 1:

The system implements periodic listening windows where the wake-word engine is activated only during specific time intervals (e.g., when media playback is occurring or user interaction is detected), rather than continuous monitoring. This periodic activation maintains responsiveness to legitimate commands while significantly reducing false positives from ambient audio during non-listening periods.

Inventive Principle:

Principle #19Periodic action

Solution Approach 2:

The listening state of the wake-word engine is made dynamic rather than static. The system adjusts its monitoring behavior based on current playback conditions, media state, and user interaction patterns, transitioning between active listening and inactive states to optimize both responsiveness and accuracy in real-time.

Inventive Principle:

Principle #15Dynamics

2Adaptability or versatility

If wake-word engines continuously monitor ambient audio, then voice command availability is improved, but resource consumption increases

Engineering Contradiction:

Improvevoice command availabilityVSAvoidresource consumption

Core Design Contradiction:

Adaptability or versatilityVSUse of energy by moving object

Solution Approach 1:

The system implements periodic listening windows where the wake-word engine is activated only during specific time intervals (e.g., when media playback is occurring or user interaction is detected), rather than continuous monitoring. This periodic activation maintains responsiveness to legitimate commands while significantly reducing false positives from ambient audio during non-listening periods.

Inventive Principle:

Principle #19Periodic action

Solution Approach 2:

The system determines media playback events in advance (such as track changes, playback state transitions) and proactively opens listening windows before user commands are likely to occur. This preliminary action ensures voice commands are available when needed without requiring continuous monitoring, thereby reducing resource consumption while maintaining availability.

Inventive Principle:

Principle #10Preliminary action

3Difficulty of detecting and measuring

If wake-word engines continuously monitor ambient audio, then command detection capability is improved, but interruptions to media playback increase

Engineering Contradiction:

Improvecommand detection capabilityVSAvoidmedia playback continuity

Core Design Contradiction:

Difficulty of detecting and measuringVSStability of the object's composition

Solution Approach 1:

The system implements periodic listening windows where the wake-word engine is activated only during specific time intervals (e.g., when media playback is occurring or user interaction is detected), rather than continuous monitoring. This periodic activation maintains responsiveness to legitimate commands while significantly reducing false positives from ambient audio during non-listening periods.

Inventive Principle:

Principle #19Periodic action

Solution Approach 2:

The system uses feedback from media playback state (such as track progress, playback completion, user interaction patterns) to dynamically adjust listening window timing and duration. This feedback mechanism ensures commands are detected when most likely to occur while minimizing interruptions to media playback flow.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20250166625A1Input detection windowing

Publication Date: 2025.05.22 SONOS INC

AI summary

A device, such as Network Microphone Device or a playback device, detecting an event associated with the device or a system comprising the device. In response, an input detection window is opened for a given time period. During the given time period the device is arranged to receive an input sound data stream representing sound detected by a microphone. The input sound data stream is analyzed for a plurality of keywords and/or a wake-word for a Voice Assistant Service (VAS) and, based on the analysis, it is determined that the input sound data stream includes voice input data comprising a keyword or a wake-word for a VAS. In response, the device takes appropriate action such as causing the media playback system to perform a command corresponding to the keyword or sending at least part of the input sound data stream to the VAS.