Media Content Analysis Using Speech and OCR for Voice Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Voice recognition systems struggle to accurately identify media content due to the dynamic and varied vocabulary used in describing media assets, leading to difficulties in user searches for relevant media content.
Innovation Solution
A media search system that processes advertisements to extract keywords and phrases associated with media content, using speech recognition and optical character recognition to enrich metadata, and incorporates user voice inputs to improve voice-based search capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a voice recognition system uses a fixed vocabulary for media content, then the system structure remains simple, but the system cannot adapt to dynamic and varied vocabulary used in describing media assets
Solution Approach 1:
The patent implements a dynamic vocabulary system where the media search system continuously learns and updates keywords from multiple sources including advertisements, social media, and user interactions. The vocabulary is not static but evolves over time to match changing media content descriptions, resolving the contradiction between adaptability and complexity by making the system dynamically adjustable rather than fixed
Solution Approach 2:
The system performs preliminary actions by proactively collecting and processing keywords from advertisements and social media before users perform searches. This advance preparation of vocabulary data allows the system to be ready with updated keywords when search queries occur, improving adaptability without adding complexity to the core search operation
2Measurement precision
If the media search system uses extensive keyword extraction from multiple sources, then search accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The system extracts and processes keywords from advertisements and social media in advance, before users perform searches. This preliminary keyword extraction and indexing allows the system to have search-ready vocabulary data prepared beforehand, improving search accuracy while avoiding time-consuming processing during actual user search operations
Solution Approach 2:
The patent divides the keyword extraction process into separate modular components: advertisement processing, social media monitoring, user interaction analysis, and search query processing. Each component operates independently and can be optimized separately, allowing accurate keyword extraction without proportionally increasing overall processing time
3Adaptability or versatility
If the system processes and stores keywords from advertisements and social media, then the vocabulary coverage expands, but data management complexity increases
Solution Approach 1:
The patent implements a universal keyword database that serves multiple functions: it stores keywords from advertisements, social media, user interactions, and internal search queries. This single multi-functional database structure allows the system to manage diverse vocabulary sources without proportionally increasing data management complexity, as the same infrastructure handles all keyword types
Data Source
AI summary
Methods and apparatus for improving accuracy in media content searches are described. A video stream may be analyzed to identify one or more words associated with a media content item. The one or more words may be used to fulfill a search request and causing output of the media content item.