Local Keyword Detection via Segmented NLU Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice-assisted media playback systems face challenges in accurately detecting voice commands without false positives, particularly in noisy environments, and often require cloud-based processing, which raises privacy concerns and may delay responses due to the need for remote data transmission.
Innovation Solution
Implementing a network microphone device with a command-keyword engine that detects specific keywords locally, combined with a local natural language unit, to process voice inputs and perform actions only when certain conditions are met, reducing false positives and maintaining user privacy by minimizing cloud interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If cloud-based processing is used for voice command detection, then processing power and keyword library size can be increased, but user privacy is compromised and response time increases due to data transmission
Solution Approach 1:
The system segments the voice processing function into two parts: local keyword spotting (detecting wake words and commands) and cloud-based verification (confirming ambiguous keywords). This segmentation allows the system to maintain a comprehensive keyword library in the cloud while performing basic detection locally, thus preserving privacy for clear commands while verifying uncertain ones remotely.
Solution Approach 2:
The system introduces an intermediary verification mechanism that acts as a mediator between local processing and cloud processing. Only keywords that meet certain uncertainty criteria are transmitted to the cloud for verification, while confident local detections are processed immediately without cloud intervention, thus minimizing data transmission while maintaining accuracy.
2Adaptability or versatility
If cloud-based processing is used for voice command detection, then processing power can be increased, but response time increases due to remote data transmission
Solution Approach 1:
The system segments processing into immediate local responses and deferred cloud verification. Common wake words and clear commands are processed instantly locally, providing fast response times, while only ambiguous cases are sent to the cloud for later verification, thus minimizing the impact of cloud processing delays on overall system responsiveness.
Solution Approach 2:
The system performs preliminary local keyword spotting and confidence assessment before cloud verification. By pre-filtering and identifying only uncertain keywords for cloud processing, the system reduces the volume of data requiring remote transmission and enables faster overall processing through this two-stage approach.
3Loss of time
If local processing is used for voice commands, then response time and privacy are improved, but false positives increase in noisy environments
Solution Approach 1:
The system implements a feedback mechanism where cloud verification results are used to adjust local processing behavior. When the cloud verifies a local detection as incorrect, this feedback is used to refine local keyword spotting parameters and confidence thresholds, progressively reducing false positives in noisy environments through continuous learning from verification outcomes.
Solution Approach 2:
The cloud verification service acts as an intermediary that filters out false positives detected locally. By having the cloud verify uncertain local detections, the system eliminates false positives that local processing alone cannot distinguish, thus improving reliability while maintaining the speed benefits of local processing for clear cases.
4Reliability
If local keyword detection is implemented, then user privacy is protected and response time is reduced, but measurement precision decreases in noisy environments
Solution Approach 1:
The system segments keyword detection into high-confidence local detections and low-confidence cloud verifications. By separating these cases based on confidence thresholds, the system maintains high privacy protection for clear local detections while using cloud resources to improve accuracy for ambiguous cases, thus achieving both privacy and precision goals simultaneously.
Solution Approach 2:
The cloud verification service serves as an intermediary that improves measurement precision for uncertain cases without compromising privacy for confident local detections. This intermediary layer provides accuracy enhancement only where needed, maintaining the privacy benefits of local processing while correcting precision deficiencies in noisy environments.
Data Source
AI summary
In one aspect, a playback device includes at least one microphone configured to detect a voice input and generate sound input data. The playback device detects a first command keyword in the detected sound and, in response, makes a first determination, via a first local natural language unit (NLU), whether the input sound data includes at least one keyword within a first predetermined library of keywords. The playback device receives an indication of a second determination made by a second NLU that the input sound data includes at least one keyword from a second predetermined library of keywords. The playback device compares the results of the first determination and the second determination and, based on the comparison, foregoes further processing of the input sound data.