forum | IKS - Details

forum | IKS: Vortrag von Dr.-Ing. habil. Gerald Enzner

Institut für Kommunikationsakustik, Ruhr-Universität Bochum
Multichannel Sound Acquisition and Adaptive Enhancement
Dienstag, 28. Juni 2016
17:00 Uhr
Hörsaal 4G

 

Im Rahmen der neuen Veranstaltungsreihe lädt das Institut für Kommunikationssysteme zum letzten Vortrag des Sommersemesters ein. Herr Dr.‐Ing. habil. Gerald Enzner vom Institut für Kommunikationsakustik der Ruhr‐Universität Bochum wird zum Thema "Multichannel Sound Acquisition and Adaptive Enhancement" referieren.

Gegenstand seines Vortrags sind mehrkanalige Systeme zur Verbesserung von Audiosignalen. Der damit verbundene Einsatz von mehreren Mikrofonen führt sowohl in der Hardware‐Konfiguration als auch bei den eingesetzten Verfahren zu mehr Freiheitsgraden. Die zusätzlich gewonnene Information lässt sich zu einer verbesserten Störbefreiung oder Enthallung nutzen. Dabei spielt u.a. die adaptive Bayessche Modellierung eine entscheidende Rolle. Die grundsätzlichen theoretischen Betrachtungen werden abschließend anhand von Beispielen aus der Praxis veranschaulicht.

Im Anschluss an das Fachprogramm haben die Gäste die Möglichkeit, bei einem Imbiss mit dem Vortragenden zu diskutieren.

Abstract:
Multichannel sound enhancement uses multiple microphones and their information diversity to reconstruct cleaner sound from originally noisy or reverberant recordings. The availability of multiple microphones however generates many possibilities regarding the actual microphone-array configuration and the corresponding signal processing. This talk advocates the utility of adaptive Bayesian modeling as a means to resolve the actual acoustic configuration from the noisy signals while performing the sound enhancement. In applications with uncertainty of more than one acoustic feature, e.g., source, channel, or observation noise, we refer to blind adaptive signal processing in order to express the difficulty related to multiple uncertainties. "Blind multichannel equalization and identification" (BENCH) for acoustic dereverberation is then reported with some detail. Results of the sound enhancement are depicted for speech listening and speech recognition applications.

Biography: 
Gerald Enzner received the Dipl.-Ing. degree from University of Erlangen-Nuremberg, Germany, in 2000, and the Dr.-Ing. degree from RWTH Aachen University, Germany, in 2006, both in electrical engineering and information technology. Since 2007, he has been working at Ruhr University Bochum, Germany, as a Principal Investigator in the area of adaptive and acoustic signal processing. Also, Gerald Enzner has been a visiting researcher to further academic and industrial organizations, for instance, Siemens Corporate Research, Princeton, NJ, the TU Eindhoven, Netherlands, the Nokia Research Center and Helsinki University of Technology, Finland, and Starkey Hearing Technologies, Eden Prairie, MN. In the 2009–2011 term, he was a member of the Global Young Faculty of the University Alliance Metropolis Ruhr (UAMR) and the Mercator Foundation. In 2013, he has been appointed to faculty level via Habilitation in the Department of Electrical Engineering and Information Technology at Ruhr-University Bochum.

zurück