Smart Notification System for Video Conferencing Keyword Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
During online video conferencing sessions, users often miss important information due to lack of engagement, as they may not be actively paying attention and miss key discussions related to registered keywords.
Innovation Solution
Implementing a smart notification system on client devices that uses a notification gateway and a low-power speech recognition accelerator to locally analyze audio data from video conferencing applications, identifying registered keywords and generating real-time notifications to re-engage users without sending audio data remotely, thus ensuring privacy and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If users actively pay attention to the entire video conferencing session, then they can capture all important information, but their engagement level and ability to focus on other tasks decreases
Solution Approach 1:
The system performs automatic keyword detection and notification generation without requiring continuous user attention. The speech recognition accelerator autonomously monitors the audio stream, identifies registered keywords, and triggers notifications, allowing the system to serve itself rather than requiring active user monitoring of the entire session.
Solution Approach 2:
The system provides immediate feedback to users when registered keywords are detected in the audio stream. This feedback mechanism (notifications) informs users of important information moments, enabling them to re-engage at critical points without needing to maintain constant attention throughout the entire session.
2Measurement precision
If audio data is sent remotely for analysis, then comprehensive keyword detection can be achieved, but privacy concerns and data transmission time increase
Solution Approach 1:
The speech recognition accelerator extracts and processes only the essential audio features locally on the client device. Instead of transmitting entire audio streams remotely, the system extracts keyword-related information locally, sending only necessary data for notification generation, thereby maintaining privacy while achieving detection accuracy.
Solution Approach 2:
The speech recognition accelerator acts as an intermediary component between the audio input and remote servers. It processes audio data locally first, performing preliminary keyword detection and filtering, thereby reducing the need for comprehensive remote analysis and minimizing privacy-exposing data transmissions.
3Loss of information
If traditional notification systems are used without keyword detection, then system complexity remains low, but users miss important information during their inattention periods
Solution Approach 1:
The system changes the operational parameters of the notification mechanism by integrating speech recognition capabilities. Instead of simple timestamp-based or manual notifications, the system monitors audio stream parameters (keyword presence) to trigger notifications dynamically, improving information capture while managing complexity through parameter-based control.
Solution Approach 2:
Users pre-register keywords before the video conferencing session begins. This preliminary action allows the system to have detection targets ready in advance, enabling immediate keyword-based notification triggering without requiring complex real-time decision-making about what constitutes important information during the session.
4Speed
If speech recognition is performed remotely, then processing power requirements for client devices remain low, but notification speed and user response time decrease
Solution Approach 1:
The speech recognition task is segmented into two parts: local keyword detection using the speech recognition accelerator for immediate notification triggering, and optional remote verification for comprehensive analysis. This segmentation enables fast local responses for urgent notifications while using remote resources for less time-critical processing, balancing speed and energy consumption.
Data Source
AI summary
This disclosure describes systems, methods, and devices related to smart notifications for online collaboration applications. A method may include receiving, by at least one processor of a device, audio data from an audio stream presented using a video application of the device; identifying, by the at least one processor, a keyword for which to search in the audio data; determining, by the at least one processor, that the audio data includes a representation of the keyword; generating, by the at least one processor, based on the determination that the audio data includes the representation of the keyword, a notification indicating that the keyword was identified in the audio data; and causing presentation, by the at least one processor, of the notification using the device.