RĂ¼diger Hoffmann and Matthias Wolff


Speech and many other acoustic signals show a hierarchical structure which has to be considered if systems for speech and audio processing are developed. Because systems for speech recognition and for speech synthesis follow the same hierarchy, a unified approach (UASR) was proposed in the year 2000, which was implemented during the following decade. The general application of Finite State Transducers (FST) results in a very efficient technology at all symbolic levels of the hierarchy. UASR proved to be successful not only for speech processing but also for many applications to other biological, technical, or musical signals, resp. At the same time, the idea of cognitive dynamic systems became popular mainly due to the work of S. Haykin. It is very promising to expand the UASR system to a hierarchical cognitive dynamic system, combining the hierarchical structure of UASR with the approach of cognitive systems, which is mainly elaborated for the so-called cognitive radio so far. The target system, the structure of which was defined now, will perform intelligent processing of speech and other signals.


intelligent signal processing, hierarchical systems, cognitive systems, acoustic pattern recognition, audio processing