Acoustic Pattern Filtering to Prevent Voice Assistant Triggers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The unpredictability of triggering personal assistant applications on electronic devices due to the presence of keywords in audio data, which can occur unintentionally and are often inaudible, poses a challenge for users, especially in shared environments.
Innovation Solution
A method to detect and correct acoustic patterns in audio streams by applying corrective actions such as jamming, omitting, or scrambling the audio data to prevent the invocation of personal assistant applications, using frequency and amplitude modifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If audio data is played through output devices, then users can enjoy audio content, but personal assistant applications may be unintentionally triggered by keywords in the audio
Solution Approach 1:
The system performs preliminary detection of acoustic patterns in the audio stream before playback. By identifying keywords and acoustic signatures that could trigger personal assistant applications in advance, the system can prepare corrective actions (such as jamming signals or audio scrambling) to prevent unintended activations during playback.
Solution Approach 2:
The system applies preliminary anti-action by detecting potential triggering keywords and counteracting them with corrective actions. When a keyword is detected in the audio stream, the system generates jamming signals or scrambles the audio in that specific time frame to prevent the personal assistant application from being activated, while minimizing impact on the overall audio quality.
2Reliability
If audio scrambling or jamming is applied to prevent keyword triggering, then unintended activations are reduced, but audio quality may be degraded
Solution Approach 1:
The system applies local quality by implementing corrective actions only in specific time frames where keywords are detected, rather than degrading the entire audio stream. The jamming signals or audio scrambling are localized to the precise time frames containing triggering keywords, leaving the rest of the audio content unchanged and preserving overall audio quality.
Solution Approach 2:
The system uses partial action by applying corrective measures only to the extent necessary - specifically targeting the time frames containing keywords rather than the entire audio stream. This partial application of jamming or scrambling minimizes the impact on audio quality while still effectively preventing unintended activations.
3Reliability
If the system detects and modifies acoustic patterns in real-time, then triggering is prevented, but processing complexity increases
Solution Approach 1:
The system applies segmentation by dividing the audio stream into discrete time frames and analyzing each frame independently for acoustic patterns. This segmentation approach allows the system to process audio in manageable chunks, identifying keywords in specific time frames and applying corrective actions only where needed, rather than processing the entire audio stream as a single complex task.
Data Source
AI summary
According to an example, a method comprises receiving a first audio stream from an input device, detecting presence within the first audio stream of at least an acoustic pattern, executing at least one corrective action over a portion of data of the first audio stream including the acoustic pattern such that a second audio stream is obtained, and transmitting the second audio stream to an output device.