Current speech recognition engines can not provide a 100% robust detection. This means that devices such as your mobile phone, TV or PlayStation require a keyboard to type at (very) fast pace. With the SpeeG version 1 approach, we are currently trying to tackle this problem and are under the belief that in the near future one can type faster on a TV than with a traditional keyboard.

Interested students should contact us for more information.

SpeeG version 1 will hopefully be a success story, but we still have some additional ideas to optimise it. One aspect is to improve the speech recognition system without additional effort of the user. A second aspect is to support multiple users by defining a different speech model for each user. In short, the SpeeG thesis topic aims at completely removing the need of keyboard devices for efficient text input.

This thesis will be done in collaboration with the Computational Modeling Lab (CoMo). Contacts: Yann-Michaël De Hauwere, Ann Nowe.