Ingredients and Recipe for a Robust Mobile Speech-enabled Cooking Assistant for German

Ulrich Schäfer, Frederik Arnold, Simon Ostermann, Saskia Reifers

In: Ingo J. Timm, Matthias Thimm (editor). KI 2013: Advances in Artificial Intelligence. Pages 212-223 Lecture Notes in Artificial Intelligence (LNAI) 8077 ISBN 978-3-642-40941-7 Springer Berlin - Heidelberg 9/2013.


We describe the concept and implementation of Kochbot, a cooking assistant application for smartphones and tablet devices that robustly processes speech I/O and supports German recipes. Its main functions are (1) helping searching in a large recipe collection, (2) reading out loud the cooking instructions step-by-step, and (3) answering questions during cooking. Our goal was to investigate and demonstrate the use of speech assistance in a task-oriented, hands-free scenario. Furthermore, we investigate rapid domain adaptation by utilizing shallow natural language processing techniques such as part-of-speech tagging, morphological analysis and sentence boundary detection on the domain text corpus of 32,000 recipes. The system is fully implemented and scales up well with respect to the number of users and recipes.

Weitere Links

kochbot-lncs8077-preprint.pdf (pdf, 768 KB ) kochbot_ki2013.bib (bib, 723 B )

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz