Andreas Stolcke

Projects

I am involved in the following ongoing projects at SRI:

• Effective Affordable Reusable Speech-to-text (EARS) (with ICSI and U. Washington)
• Speech in Noisy Environments (SPINE)
• Prosody for Dialog Systems
• Meeting Recognition and Modeling (with ICSI)
• Mapping Meetings (with ICSI, U. Washington, and Columbia U.)
• SRI Language Modeling Toolkit

• Hidden Word-level Events in Spontaneous Speech
• Large Vocabulary Conversational Speech Recognition
• Information Extraction from Speech (SIEVE)
• Speech Prosody and Disfluencies
• Spoken Language Translation
• Consistency Modeling

I also coordinate the language modeling work at SRI's speech laboratory across projects.

Apart from my work at SRI, I have also participated in the Summer Speech Workshops at Johns Hopkins University's Center for Language and Speech Processing.

• The 1995 Language Modeling Workshop, where I participated in a research project on "Fast Sparse Data Training/Portability".
• The 1996 Workshop on Innovative Techniques for Large Vocabulary Conversational Speech Recognition, where I led a group investigating "Dependency Language Modeling". (Also see the group home page .)
• The 1997 Workshop on Innovative Techniques for Large Vocabulary Conversational Speech Recognition, where I was part of a group investigating Discourse Language Modeling.

Previously, for my thesis work, I created another object-oriented software system (in Lisp/CLOS) to experiment with probabilistic grammars and learning algorithms. That system, called BOOGIE, is also freely available for research use.

Previous projects are described on my home page at ICSI where I did my graduate work.

Andreas Stolcke's Home Page
Speech Group Home Page
SRI Home Page