We present a system for Continuous Speech Understanding tasks, which studies the learning of transducers for these tasks within the framework of Language Models; that is, it modelizes certain linguistic units and constrains the possible concatenation of such units. Both the learning of the acoustic models corresponding to the linguistic units, and the unit concatenation rules are inductively obtained through Grammatical Inference techniques. Some results obtained for the task of understanding Spanish numbers are reported. KEYWORDS: Continuous Speech Understanding, Language Models," Transduction, Grammatical Inference.
Cite as: Segarra, E., Garcia, P. (1991) Automatic learning of acoustic and syntactic-semantic levels in continuous speech understanding. Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991), 861-864, doi: 10.21437/Eurospeech.1991-213
@inproceedings{segarra91_eurospeech,
author={Encarna Segarra and Pedro Garcia},
title={{Automatic learning of acoustic and syntactic-semantic levels in continuous speech understanding}},
year=1991,
booktitle={Proc. 2nd European Conference on Speech Communication and Technology (Eurospeech 1991)},
pages={861--864},
doi={10.21437/Eurospeech.1991-213}
}