Audio Content Search via Text Extraction in CMS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Content management systems (CMS) face limitations in searching media files, particularly audio data, as they lack the capability to search within the content of media files, restricting effective media content management and retrieval.
Innovation Solution
A system and method that converts audio data into text, extracts specified strings with attributes, and stores media files as searchable objects within a content management repository, allowing for the retrieval of media files based on search queries within their content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional CMS stores and manages media files, then content management capability is provided, but search capability within media file content is limited or non-existent
Solution Approach 1:
The patent introduces an intermediary layer between the media file storage system and the search query processing system. This intermediary comprises a speech recognition module that converts audio data to text, and a string extraction module that identifies searchable strings from the transcribed text. This intermediary layer enables search functionality without requiring the entire media management system to be rewritten, thus improving search capability while managing system complexity through modular architecture.
Solution Approach 2:
The search functionality is segmented into separate processing stages: audio data reception, speech recognition conversion to text, string extraction from the text, attribute determination for extracted strings, and search query matching. This segmentation allows each component to be optimized independently and facilitates easier implementation and maintenance of the search capability within the CMS.
2Loss of information
If audio data is stored as media files without text conversion, then storage efficiency is maintained, but searchability of audio content is lost
Solution Approach 1:
The system performs preliminary action by converting audio data to text and extracting searchable strings during the media file ingestion process, before the actual search operations are needed. The speech recognition and string extraction are performed in advance when media files are uploaded or processed, creating an indexed representation of the audio content that can be quickly searched later. This preliminary processing reduces the overhead during search operations and ensures searchability is established early in the data lifecycle.
Solution Approach 2:
The patent creates a textual copy or representation of the audio data through speech recognition. Instead of searching the original audio files directly, the system creates a text-based copy of the audio content's information, which is then made searchable. This copying approach preserves the original audio data for storage while creating an accessible text representation that enables search functionality without duplicating the entire audio content.
3Productivity
If media files are made searchable through content extraction, then retrieval efficiency is improved, but processing time for indexing increases
Solution Approach 1:
The system implements periodic action by performing string extraction and attribute determination in batches or periodically during media file processing, rather than continuously. The speech recognition and string extraction operations are scheduled to process audio files in manageable chunks, allowing the system to maintain retrieval efficiency through indexed data while managing processing time through batched operations that can be optimized based on workload patterns.
Solution Approach 2:
The patent changes parameters such as the granularity of string extraction, the level of detail in attribute determination, and the threshold for creating searchable entries. By adjusting these parameters, the system can optimize the balance between retrieval efficiency and indexing processing time. For example, extracting only key phrases rather than all words, or determining only essential attributes, can reduce processing time while maintaining sufficient searchability for practical purposes.
Data Source
AI summary
A system comprising a client computer, a data store comprising a content management repository, a server computer coupled to the client computer by a network, the server computer comprising code for: receiving audio data; converting the audio data to text; extracting a specified string from the text as an extracted string; determining an extracted string attribute for the extracted string; storing a media file containing the audio data as a content object; configuring the content object to be searchable by the extracted string; receiving a search query from the client application; searching a plurality of managed objects based on the search query; and based on determining that the extracted string matches a search string, returning an indication of the first media file, the extracted string and the extracted string attribute in a search result.