Online Conference Anomalous Sound Mitigation via ML Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online conferencing systems face challenges in effectively mitigating anomalous sounds, such as dog barking or sirens, which can disrupt meetings, as existing solutions struggle to identify and address these issues in real-time.
Innovation Solution
A system that records streaming media during online conferences, uses machine learning to analyze feedback from participants, and applies filters or muting to mitigate anomalous sounds by crowdsourcing feedback and analyzing recorded media streams to identify and address the source of the noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing solutions are used to identify and mitigate anomalous sounds, then the system can maintain basic conference functionality, but the ability to effectively identify and address anomalous sounds in real-time deteriorates
Solution Approach 1:
The patent introduces a mediator component that collects feedback from multiple conference participants about anomalous sounds and processes this information to identify the source. This intermediary layer enables effective sound mitigation without requiring direct complex analysis from each participant, resolving the contradiction between reliability and device complexity.
Solution Approach 2:
The system implements a feedback mechanism where participants report anomalous sounds they hear, and this feedback is aggregated and analyzed to identify the source. The system then applies mitigation and provides feedback to participants about the action taken. This feedback loop enables reliable sound identification and mitigation while keeping the system manageable in complexity.
2Measurement precision
If the system records and analyzes all streaming media data to identify anomalous sounds, then the precision of sound identification improves, but the processing time and computational resources increase
Solution Approach 1:
The patent extracts only the necessary information for sound identification from the complete streaming media data. Instead of analyzing all media data in full detail, the system extracts specific audio characteristics and feedback information that are relevant to identifying anomalous sounds, thereby improving precision while reducing processing time and computational resources.
3Speed
If the system implements real-time analysis of streaming media, then the speed of anomalous sound detection improves, but the use of computational resources deteriorates
Solution Approach 1:
The system performs partial analysis of the streaming media data, focusing only on detecting anomalous sounds rather than comprehensively analyzing all audio characteristics. This partial action approach enables real-time detection speed while reducing computational resource consumption by avoiding unnecessary analysis of normal audio segments.
Data Source
AI summary
Methods, computer program products, and systems are presented. The method computer program products, and systems can include, for instance: activating a streaming media recording buffer that records streaming media of an online conference, the online conference having first second and third user online conference participant users; examining data to return an action decision, the examining data to return an action decision including examining data of the streaming media recording buffer to identify an anomalous sound represented in the recorded media stream data of the streaming media recording buffer; returning an action decision based on the examining data to return an action decision, the action decision being an action to mitigate the anomalous sound; and providing one or more output to mitigate the anomalous sound in accordance with the returned action decision.