This site has been archived on
The Community Research and Development Information Service - CORDIS
Information & Communication Technologies

Language Technologies


back to overview

Project factsheets will no longer be updated.  All information relevant to the project can be found on the CORDIS factsheet .  This is updated on a regular basis with public deliverables, etc.

ORGANIC - Self-organized recurrent neural learning for language processing

231267 - STREP


At a glance

ICT-2007.2.2 - Cognitive Systems,
Interaction, Robotics


Current speech recognition technology is based on mathematical-statistical models of language. Although these models have become extremely refined over the last decades, progress in automated speech recognition has become very slow. Human-level speech recognition seems unreachable. The ORGANIC project ventures on an altogether different route toward automated speech recognition: not starting from statistical models of language, but from models of biological neural information processing – from neurodynamical models.


ORGANIC will combine a large variety of neurodynamical mechanisms – of signal filtering, learning, short- and longterm memory, dynamical pattern recognition – into a complex neurodynamical "Engine" for speech recognition. A major scientific challenge of ORGANIC lies in the very complexity of the targeted architectures. In order to master the resulting nonlinear dynamical complexity, a special emphasis is put on mechanisms of adaptation, self-stabilization and selforganization. The overall approach is guided by the paradigm of Reservoir Computing, a biologically inspired perspective onhow arbitrary computations can be learnt and performed in
complex artificial neural networks.

The result and Impact

  • Establishing neural dynamics as viable technological alternative for speech and handwriting recognition:Current technology in these fields is data-centered, relying on a statistical analysis of the concerned speech and handwriting data. In ORGANIC, the scientific approach starts from artificial recurrent neural networks, computational neuroscience, cognitive neuroscience and nonlinear dynamics; it may justly be termed mechanism-centered. The most fundamental desired outcome of ORGANIC is to prove that its mechanism-centered approach represents a viable alternative to the data-centered one. While the project does not promise to surpass existing technology – that would be preposterous given that current technology reflects many thousands of person-years of effort –, the project aims at neural implementations whose performance reaches state-of-the-art levels. Besides performance on benchmarks, reaching this goal will be evaluated by the inclusion of neurodynamical modules into commercial products on the side of the consortium SME partner Planet intelligent systems GmbH.
  • Services to the academic community: Through a public-domain Engine which adheres to existing standards, through a benchmark repository (public as far as proprietary rights permit), through detailed public documentation of implementations in online technical reports, and through open workshops, the technologies developed in ORGANIC shall be made easily accessible to and useable for the scientific community.
  • Acceptance of the neurodynamical approach in the scientific community will be demonstrated by peer-reviewed scientific articles. ORGANIC has committed to generate 15 accepted papers in highranking journals and 30 accepted contributions to leading conferences.

Contact Person:

Name: Marta Müller






back to overview


This page is maintained by: Susan Fraser (email removed)