EBU experts working on automatic metadata extraction have published a public Request for Technology. The experts are cooperating under the umbrella of the EBU Group called 'P/SCAIE': Content Analysis-based Automatic Information Extraction.
Four areas of work
One of the key tasks of the P/SCAIE Group is to facilitate the introduction of new production tools based on the automatic extraction of information (including metadata). This data results from the analysis of audiovisual content. In the first phase, the P/SCAIE project has identified four promising areas to focus on:
1. Speech recognition
2. Semantic segmentation and shot detection
3. Audio-visual copy & repetition detection
4. Content summarisation
Requirements
A set of specific requirements has been identified for each of these four areas, and test material has been selected and provided by P/SCAIE participants. The Request for Technologies document can be downloaded here. EBU members which are not yet part of the project are encouraged to join P/SCAIE, to contribute with their experience, to share knowledge about tools, or simply to learn the basics of this exciting subject.