Live Event Summary Generation Using Audio-Intensity Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying relevant portions of live events for summaries is labor-intensive and often occurs after the event, requiring human intervention and increasing costs and time when multiple events occur simultaneously.
Innovation Solution
Automated system that analyzes multimedia streams from live events, divides them into scenes, and ranks activities based on audio signal intensities to generate summaries without manual selection, allowing real-time or post-event generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human reviewers manually identify and select relevant portions for summaries, then the accuracy and relevance of summaries is improved, but the time required and labor cost increases significantly
Solution Approach 1:
The patent replaces the manual mechanical process of human reviewers watching broadcasts and selecting clips with an automated computer-based system. The system uses audio signal processing, scene detection algorithms, and automated selection criteria to identify and compile summary clips without human intervention, thereby eliminating the time cost while maintaining summary quality through objective automated decision-making
Solution Approach 2:
The system performs self-service by automatically analyzing broadcast content, detecting scenes, measuring audio intensities, and selecting relevant portions for summaries without requiring external human input. The automated system serves itself by using the broadcast's own audio and visual data to generate summaries, eliminating the need for human reviewers to watch and select content manually
2Measurement precision
If human reviewers manually identify and select relevant portions for summaries, then the relevance and quality of summaries is improved, but the cost increases significantly
Solution Approach 1:
The patent replaces the manual mechanical process of human reviewers with an automated computer-based system that uses audio signal processing and scene detection algorithms to identify and compile summary clips, thereby eliminating labor costs while maintaining summary quality through objective automated decision-making
Solution Approach 2:
The system performs self-service by automatically analyzing broadcast content, detecting scenes, measuring audio intensities, and selecting relevant portions for summaries without requiring external human input. The automated system serves itself by using the broadcast's own audio and visual data to generate summaries, eliminating the need for human reviewers and associated labor costs
3Measurement precision
If summaries are generated after the live event concludes, then the accuracy of summary content is improved, but the timeliness and usefulness of summaries is reduced
Solution Approach 1:
The system performs preliminary action by continuously monitoring and analyzing audio signals during the live broadcast, detecting scenes and measuring audio intensities in real-time. This preliminary processing allows the system to identify and prepare summary clips while the event is still ongoing, enabling faster summary generation without compromising accuracy since the analysis is performed continuously rather than after the event concludes
Data Source
AI summary
Summaries of broadcasts of live events are generated based on intensities of audio signals captured during the live events. Streams of multimedia including video signals and audio signals are captured by one or more cameras. The video signals are processed to identify specific activities of interest (e.g., plays of a sporting event) during the live event. Audio signals captured concurrently with the video signals are processed to determine their respective intensities or other acoustic characteristics. The activities of interest are ranked based on intensities of the audio signals. A multimedia stream representing a summary of a media program and includes the highest-ranking video signals and corresponding audio signals is generated and transmitted to one or more devices of viewers.