In the present study, we defend the thesis that numerous studies of extensive spontaneous speech corpora are needed to improve the performance of Natural Language Processing systems (NLP). We consider that at the present stage of research, more attention should be paid to the variability of human discourse and the probabilistic approach, which will contribute significantly to the better intelligibility and naturalness of synthesized speech. In our contribution, we will address the question of automatic prosodic annotation of speech corpora; consider the problem of choosing appropriate units of description and of the automatic extraction and coding of these units.
Cite as: Nesterenko, I. (2004) Towards the integrations of stochastic information in speech technologies: the case of suprasegmentals. Proc. 9th Conference on Speech and Computer (SPECOM 2004), 578
@inproceedings{nesterenko04_specom,
author={Irina Nesterenko},
title={{Towards the integrations of stochastic information in speech technologies: the case of suprasegmentals}},
year=2004,
booktitle={Proc. 9th Conference on Speech and Computer (SPECOM 2004)},
pages={578}
}