Harald Romsdorfer studied Economics at the Economic University of Vienna, Austria and Electrical Engineering at the Vienna University of Technology, Austria with a focus on Computer Technology, Signal Processing, Telecommunication, and Mobile Communication. During this time, he made internships at Siemens, Ericsson, Teleste, and Austrian Aerospace. He received the Electrical Engineering degree (Diplom-Ingenieur) in 2001, for designing and implementing a VoIP-software.
From July 1998 to September 2001, he joined Philips Speech Processing in Vienna, Austria. There he was involved in the developement of the SpeechMagic™ automatic speech recognition system. At Philips, his main interests were automatic speech recognition and software architecture.
From July 2001 to January 2002, he was with the Software Competence Center Hagenberg, Austria. First as an independent consultant and since October 2001, as a member of the scientific staff. In this time, he was concerned with the application of the Rational Unified Process at partner companies.
In February 2002, Harald Romsdorfer joined as a scientific researcher ETH Zurich, Switzerland, where he started his PhD studies in February 2003. During his time at ETH Zurich, he constructed the polyglot Text-to-Speech synthesis system polySVOX, the worldwide first polyglot TTS synthesis system. He received the PhD degree in Electrical Engineering in January 2009.
His current scientific interests include software architecture, signal processing, statistical models for regression and classification (e.g., neural networks, hidden Markov models, Gaussian mixture models, support vector machines, ensemble models, etc.), and the optimal combination of rule-based and statistical models for applications in polyglot and monolingual speech synthesis, speech segmentation, and speech recognition.
A more detailed CV is available upon request.