The RAA project will yield a technically feasible and easily applicable set of tools for the automatic recognition of audio titles. Possible applications of this technology are automatic logging and identification of broadcasting content on both, radio and Internet sources. The innovation of RAA is that its recognition method is not based on technologies like watermarking, speech-recognition and rhythm-pattern-analysers, but on comparison of so called fingerprints. Those fingerprints are derived automatically by a specific information-reduction process. Before an audio title may be recognised its reference-fingerprint has to be calculated and stored in a central database. In addition the RAA project will build and collaborate with a framework of European institutions who have a common interest in applying techniques for recognition of audio titles. Thus, the collaboration will be strengthened on a European level. Finally a commercial detection and recognition service shall be launched after the project's end, by applying the set of re-useable modular tools developed within RAA.
The major objective of RAA is the development and implementation of a new, very fast and robust methodology for automatic audio recognition and analysis. The resulting applications will be able to observe real-time sources (broadcasting) as well as digital sources like the Internet. A title once registered can be identified every time it is available on one of the observed audio-sources. The resulting list of identified titles may be used for several purposes like autonomous monitoring of royalty payments, detection of copy-right violations, derivation of statistical numbers, enforcement of cultural laws and many more. Beside a basic set of re-useable modules, two audio analysis applications will be developed:
1. RadioListener: a software for real-time detection of broadcast music.
2. "www.audio.detective" a programme which autonomously searches the internet for audio sources like mpeg layer 3 files. As main exploitation line the launch of a commercial RAA-Observer Service will be prepared.
DESCRIPTION OF WORK
The organisation of the development work within RAA is based on common, object oriented software development methodologies. After a thorough user requirements study a system specification and software design phase is entered before the development of the actual software is initiated. The first milestone in the software development phase is the finishing of software beta versions. These beta versions will be evaluated by the industrial and user partners. Their modification requests will be considered for he instalment of the final versions. As a parallel activity the consortium will carry out a number of dissemination activities. These activities shall help to get in contact with other related efforts in order to get a wide acceptance and consciousness f the RAA-project. A commercial exploitation of the components and applications developed is prepared during the project. The main-exploitation line will end-up with the establishment of an audio-observer service, which can be used by the industry, after successful termination of RAA as a research-project.
The project work is carried out in five work packages:
WP1: Project Management This workpackage is responsible for all aspects of project management.
WP2: Architecture Within this workpackage the entire user requirement analysis and the software design are carried out.
WP3: Module development In this workpackage the core RAA audio recognition methodology will be developed. Moreover the workpackage contains all successive software development efforts.
WP4: System Verification Workpackage 4 is concerned with the validation of the intermediate and final development results. Moreover, this package is responsible for the collection of an appropriate set of test-data.
WP5: Exploitation Finally all issues of exploitation and dissemination are carried out in this package.
The implementation of RAA followed a modular concept, where the following main components have been developed. AudioDNA extractor: basic library for creation of AudioDNA Acquisition component: for ingestion of new titles including their AudioDNA Observation component: for monitoring of sound signals Content Analysis component; for comparison of AudioDNA with the AudioDNA database AudioDNA database: central SQL-database for storage of titles and AudioDNA RAA Editor: application for manual play-list completion All components have been developed platform independent, i.e. the are available both for Microsoft Windows and Linux/Unix systems. The only exception is the RAA Editor, which has an extensive graphical user interface and Windows was chosen as target platform. With these components several turnkey-solutions for different purposes can be composed. Those pre-build solutions are considered as project results; whereas the components are summarised with the term „Basic modules“.
Funding SchemeCSC - Cost-sharing contracts
1217 GP Hilversum