Timo Baumann

Prof. Dr. Timo Baumann, Diplom-Informatiker

Professor in Artificial Intelligence and Natural Language Processing at the School for CS and Maths at OTH Regensburg.

I research spoken language, in particular in interactive and multi-modal scenarios, such as in spoken dialogue systems. My particular interests lie in responsiveness in interaction and I have been one of the main drivers of incremental spoken dialogue systems. Beyond that, I create and use corpora of spoken language, such as the Spoken Wikipedia Corpus. I like difficult problems, which is why I work on prosody, and even prosody of post-modern poetry, and more recently multi-modal lip-synchronous dubbing. Feel free to have a look at my recent projects.

Recent News:

2022-12: Christian and I were invited to write an popular scientific article on automatic lip-synching (in German) (translation to English)
2022-10: Hear me chat on the Science Bench at Regensburg's Nacht.Schafft.Wissen.
2022-08: Congratulations Shravan, Christian and Debjoy! Our detailed analysis of SyncNet for audio-visual translation will appear at ICMI!
2022-07: Congratulations go out to Christian Schuler and Dominik Hauser, who's DDLitLab project on an experiment toolkit for video manipulation and human perception evaluations was accepted. I hope that I can be a good mentor for the project which is a follow-up from Christian's Bachelor's thesis.
2022-07: My colleague Sebastian Fischer and I have been awarded the OTH Regensburg prize "InnovativeLehre@OTH", nice! We'll work with student assistants to advance software-engineering self tests (that go beyond checking the functionality but also check for coding style) into Moodle and our programming 101 classes.
2022-06: Congratulations to Matthias Melzer, who's work (joint with Jan Dünnweber) on interpreting smart home data with NLP methods is getting published at IEEE SmartIoT 2022.
2022-04: Congratulations Debjoy and Shravan! Our paper on a multi-modal corpus of 16 years of Angela Merkel podcasts will appear at LREC (and is already available on arXiv). → Saha et al. (2022)
older news

2022-03: Congratulations to Anna Windbühler, who's MSc-work in ML was not only very thorough and successful but also leads to a publication in the conference on Engineering in Medicine and Biology (EMBC). → Windb\"uhler et al. (2022)
2022-02: Welcome OTH Regensburg! You can now reach me at firstname.lastnameÄToth-regensburg.de.
2021-12: I have accepted an offer from OTH Regensburg to become their new professor for AI and NLP starting in February. Hooray!
2021-11: I'll give a keynote at ESSV'22. I'm very much looking forward to meeting you in ~~Sønderborg~~ the virtual world!
2021-09: Ashutosh and my paper in which we evaluate heuristics for dubbing has been accepted at CHR'21. Yay and see you in November! → Baumann & Saboo (2021)
2021-08: Our summary paper about the SMOOTH robot has been accepted in Frontiers in Robotics and AI. → Krüger et al. (2021)
2021-06: Our Interspeech demo on subtitling integration into BigBlueButton was accepted. Yay and congratulations, Robert! → Geislinger et al. (2021)
2021-04: congrats Aditya and Debjoy for securing DAAD-WISE funding! We'll be working, together with returning intern Shravan and students from UHH, on TTS for automated dubbing this summer!
2021-03: The book stemming from the Speech quality special session at Interspeech 2017 is finally available and includes my chapter. → Baumann (2021)
2021-02: Our SMOOTH submission to HRI's late-breaking reports was accepted ... and got nominated for the award. Yay! → Fischer et al. (2021)
2021-01: Volkswagenstiftung continues to fund our Rhythmicalizer project. Yay!
2020-12: our paper on subtitling for lecture2go has been accepted for ESSV. Congrats, Benjamin, Robert and Irina! → Milde et al. (2021)
2020-11: Our COLING submission on MT in on/off-screen dubbing has been accepted. Great work, Alina, Shravan, and Supratik! → Karakanta et al. (2020)
2020-10: I've been appointed acting professor for machine learning for the winter term (and probably longer). Thanks for trusting in me, @fbihh!
2020-09: congrats to our summer interns, who've got their first paper accepted at the Deep Video Understanding workshop, great work! → Nayak et al. (2020), Video available!
2020-08: thanks for electing me as member (starting in fall) of the UHH informatics department's faculty council
2020-07: we've got funding to add automatic subtitling to Lecture2Go; Benjamin and two students will be working on improving students' ability to use and search lecture videos during #corona.
2020-06: our GermEval Shared Task on inferring Cognitive and Motivational style from text was a huge success. Thanks to Dirk and all participants! → Johannßen et al. (2020)
2020-06: find Ingo's and my contribution on prosodic differences in how you address your smart speaker at Mensch und Computer!
2020-05: welcome Shravan and Supratik, the two interns working on automated dubbing and video analysis this summer.
2020-03: both our/my Speech Prosody submissions made it. Hope to see you for the conference!
even older news

To reach me, try:

mail: timo.baumannÄŦoth-regensburg.de
for issues related to UHH: baumannÄŦinformatik.uni-hamburg.de
skype: timobaumann
tel: I'm in home office; please send an e-mail.
snail mail:

   OTH Regensburg, Fakultät für Informatik und Mathematik
   Prof. Dr. Timo Baumann
   Galgenbergstraße 32
   93053 Regensburg

consultation hours: please make an appointment by e-mail (see above)

I do speech science and technology consulting. Feel free to contact me at mail@firstnamelastnamewithoutseparation.de .

Former appointments:

Acting professor for machine learning at the Department of Informatics at Universität Hamburg, Systems Scientist at the Language Technologies Institute, Carnegie Mellon University; researcher and instructor in speech processing at Universität Hamburg (LT and NatS); research assistant at Bielefeld and Potsdam Universities before that. Various short-term appointments and free-lancing work with industry partners.

Internal

Edit | History | Recent Changes (all) | Search

Page last modified on January 25, 2023, at 05:05 PM