The objective of the project is to develop artificial intelligence-based tools supporting analysis of audiovisual content obtained from open sources. These tools will enable automatic acquisition, archiving and processing of content transmitted in various information channels, including foreign language channels, and published on the Internet. The developed solutions will be implemented in the EMMA system which supports activities in the area of media monitoring. The problem with such activities is not so much the acquisition of information, but its volume and variety. Manual monitoring of information published in numerous channels on an ongoing basis requires a large team of analysts which is costly and time-consuming.
EMMA will enable automatic acquisition of audiovisual content and its metadata from radio, television and Internet channels, including Polish-language and foreign-language content, and automatic extraction of information from video and audio. The result of audio processing will be speech transcripts, while for images it will be content of so-called tickers and other text elements presented in the video. Semantic analysis of this data will provide additional information which will be used to enhance content search query formulation.
All of the above mentioned functionalities will make it possible to monitor news feeds on an ongoing basis and efficiently search through the archival content without a need for time-consuming and computationally expensive content processing in response to each query. As part of the project, tools for analyzing content in a few selected foreign languages will be implemented, with the possibility to further expand this set in the future.
Project is financed by the National Center for Research and Development under the program Development of modern breakthrough technologies for national security and defense under the code name “SZAFIR” – Call No. 3/SZAFIR/2021
Project value 9 833 081,00 zl
