Media Content Identification via Audio Segmentation and Keyword Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio fingerprinting technologies face challenges in accurately identifying media content using spoken natural language queries, particularly in disambiguating results when background audio is present, as they struggle to differentiate between various types of media content based on limited keyword recognition.
Innovation Solution
A system that processes spoken natural language queries and environmental audio data to identify media content by separating utterances from background noise, transcribing the query, mapping keywords to specific content types, and using an audio fingerprinting engine to recognize content items matching the determined type, thereby providing accurate media content identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If audio fingerprinting is used to identify media content from spoken queries, then content recognition capability is improved, but accuracy deteriorates when background audio is present
Solution Approach 1:
The system segments the audio signal into two distinct components: the user's spoken query and the background environmental audio. By separating these signals, the system can process the query independently from background interference, thereby maintaining identification accuracy even when background audio is present.
Solution Approach 2:
The system extracts the spoken query portion from the mixed audio signal containing both query and background audio. This extraction allows the audio fingerprinting engine to focus solely on the relevant query content, eliminating the detrimental effect of background audio on identification accuracy.
2Device complexity
If keyword recognition is used to determine content type, then query processing is simplified, but disambiguation capability deteriorates
Solution Approach 1:
The system introduces an intermediary classification layer that maps identified keywords to specific content types (e.g., TV show, movie, music, podcast). This intermediary step bridges the gap between simple keyword recognition and accurate content identification, preserving disambiguation capability while maintaining processing simplicity.
Solution Approach 2:
The system changes the parameter space by transitioning from direct keyword-to-content matching to a two-stage process: keyword-to-content-type mapping, then content-type-to-content filtering. This parameter transformation enables better disambiguation while keeping the overall process manageable.
3Measurement precision
If environmental audio data is processed to separate utterances, then query accuracy is improved, but processing time increases
Solution Approach 1:
The system performs preliminary separation of the user's utterance from environmental audio data before transcription. By pre-processing the audio to isolate the query portion, the system improves transcription accuracy while limiting the time increase to only the separation step, rather than the entire processing pipeline.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for receiving (i) audio data that encodes a spoken natural language query, and (ii) environmental audio data, obtaining a transcription of the spoken natural language query, determining a particular content type associated with one or more keywords in the transcription, providing at least a portion of the environmental audio data to a content recognition engine, and identifying a content item that has been output by the content recognition engine, and that matches the particular content type.