MANGAN

Project description

Requirements

The availability of spoken audio content is increasing rapidly: audio books, audio plays, but also podcasts, radio reports, spoken content in print media and, last but not least, film content from talk shows to documentaries and feature films are based on spoken audio content.

Challenge

The content must be made accessible to users and, above all, found by them. This requires simple and precise extraction of metadata in order to integrate the content effectively and efficiently into the value chain. On the other hand, powerful and intuitive search approaches are playing an increasingly important role - especially against the backdrop of an increasing individualisation of consumption.

Background of an increasing individualisation of consumption.

Project goals

The goals of the project are two central innovation steps in the area of spoken audio content:

• Methods for automated metadata extraction will be developed and applied, indexing e.g. the content of a spoken audio file, but also other aspects such as the voice pitch and age of the speakers.
• Based on this metadata, an intuitive search will be developed and tested. Users are to be given targeted, individual access to spoken content on the basis of the MIC profile (Mood, Intent, Context) - a search procedure based on moods, intentions and contexts. The project makes audio content immediately more attractive and visible and thus forms an important building block for the independence of smaller players in a market dominated by international IT corporations - and has a positive radiance for the entire industry.

Project funding

This project is funded by the BMBF as part of the kmuInnovativ programme under FKZ 01IS22011.