I'm interested in various speech related research. Currently, I'm devoting myself to the KAIMAN project, working on automatic speaker verification/identification (SID) with a focus on difficult acoustic conditions. These include room reverberations as captured by distant microphones or artifacts and noise resulting from encoding and radio transmission. I'm further interested in producing fast and efficient implementations to allow real-time applications with limited resources.

Apart from speaker identification, I'm interested in the following speech related topics.

• Automatic speech recognition (ASR), i.e., recognize what was said (in contrast to who said it). My focus is on ASR in difficult acoustic conditions such as room reverberations captured by distant microphones. I'm also porting the LME's speech recognition system to Java, introducing scalable parallel computing for faster processing.
• Automatic speech summarization, i.e., the automatic generation of summaries from spoken language. It turns out that a combination of statistical methods and user interaction helps best to generate satisfactory summaries.
• Computer assisted learning. It turns out that students apply certain learning strategies which can be modeled using hidden Markov models. The online classification of the applied strategy allows to assist the learner with adequate aids for his or her way of learning.

For an overview of my speech related work see my publications list.