Daimler-Chrysler Research and Technology
Speech Understanding Department, ULM, DE


Experience and competence

DaimlerChrysler is one of the largest German companies. It's core business is automotive production but it has also been involved with information technology for many years. A growing area of interest is speech and language technology for applications in very different fields, ranging from voice operated mobile telephones to speech dialogue systems for call center automation.
Research in the field of speech understanding systems is carried out at the research centre in Ulm. The relevant topics addressed are man-machine-interfaces, communication technology, speech and language understanding and image understanding. The department of speech understanding systems is active in the whole field of speech and language processing. Work is done in speech coding, special signal processing of acoustic speech signals like noise compensation, speech recognition and synthesis, linguistics, dialogue management and speech and language based knowledge processing. Prototypical application systems are realized together with the relevant application departments of the DaimlerChrysler concern. An additional important field is research in statistical signal processing including neural networks and knowledge based systems.
DaimlerChrysler has developed the first German speech recognizers called ADES 1 to 3. These systems were either speaker dependent or speaker independent isolated word recognizers. One of the first recognizers SPREIN was used in an IVR system for the European Commissions ECHO information system to provide telephonic information about projects.
Later on the first speaker adaptive and speaker independent continuous word recognizers were developed. Another relevant project was the development of the first German automatic railway information systems, providing all railway information from the network of the Germand Railway. Further prototyping systems have been realized for warehouses and other IVR systems. The latest products developed by DaimlerChrysler are voice operation for mobile telephone and entertainment in the car and telephone speech recognizers for universal applications.

Role in the Project

General robustness of recognition is the main focus of much current research at DaimlerChrysler, who are therefore well-placed to provide the industrial edge to HOARSE and lead the application work of theme 4.

Research Linkages

Within the HOARSE programme the closest links will be with Sheffield, IDIAP, and Bochum.

Principal research personnel, expertise and involvement

  • Dr. Klaus Linhard, lecturer in adaptive signal processing, expertise in auditory processing, acoustics, digital signal processing, speech recognition, speech enhancement. Involvement 6 MM
  • Dr. Udo Haiber, scientist, expertise in speech recognition, speaker adaptation. Involvement 6 MM.


    Recent Publications

  • Klaus Linhard, Tim Haulick: Noise Subtraction With Parametric Recursive Gain Curves, Eurospeech 1999
  • Tim Haulick, Klaus Linhard, Peter Schrögmeier: Residual Noise Suppression Using Psychoacoustic Criteria, Eurospeech 1997