Electronic Device Speech-Request Thresholding for False Triggers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic devices misrecognize trigger words due to ambient sounds, leading to simultaneous requests for speech recognition that overwhelm servers, causing system overload.
Innovation Solution
An electronic apparatus with a processor that receives speech recognition requests, identifies request numbers within defined time sections, adjusts thresholds based on historical data, and prevents processing requests exceeding the threshold to avoid misrecognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the client device uses a start command or trigger word to detect user utterance, then the speech recognition service can be activated, but the system may misrecognize ambient sounds as trigger words causing false requests
Solution Approach 1:
The server performs preliminary analysis of the audio signal by examining energy distribution across multiple frequency bands before triggering speech recognition. This preliminary frequency domain analysis distinguishes genuine trigger words from ambient sounds by checking if the energy concentration pattern matches the expected trigger word profile, thereby reducing false recognition before full speech processing occurs
Solution Approach 2:
The system changes the detection parameters by transitioning from time-domain amplitude analysis to frequency-domain energy distribution analysis. By examining the spectral characteristics and energy concentration in different frequency bands, the system can differentiate between actual trigger words and ambient noise, improving detection accuracy while maintaining responsiveness
2Productivity
If multiple client devices simultaneously detect the same trigger word and send requests to the server, then each device can process user commands, but the server becomes overloaded with redundant requests
Solution Approach 1:
The server merges processing responsibilities by implementing a coordinator device that consolidates speech recognition requests from multiple client devices. When the coordinator detects that multiple devices are processing the same audio signal or triggering simultaneous requests, it consolidates these into a single processing task, reducing server load while ensuring that all devices receive the necessary recognition results
Solution Approach 2:
The system uses copying by having the coordinator device create and distribute a single speech recognition result to multiple client devices simultaneously. Instead of each device independently processing the same audio signal and generating separate requests, the coordinator performs the recognition once and copies the result to all relevant devices, eliminating redundant processing while maintaining service availability for all clients
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An electronic apparatus and a controlling method thereof are provided. The electronic apparatus includes a communication interface and a processor configured to: receive one or more speech recognition requests from at least one of a plurality of external apparatuses, each of the one or more speech recognition requests including a start command recognized by the at least one of the plurality of external apparatuses; identify whether a number of the one or more speech recognition requests within a time section exceeds a predetermined threshold; control speech recognition to be performed based on the number of the one or more speech recognition requests not exceeding the predetermined threshold; and transmit a result of the speech recognition to the at least one of the plurality of external apparatuses.