Adaptive Resonance Theory Neural Network for Phoneme Perception and Production
Authors
Marius CRISAN
Corresponding Author
Marius CRISAN
Available Online December 2019.
- DOI
- 10.2991/mmsta-19.2019.45How to use a DOI?
- Keywords
- adaptive resonance theory; speech perception and production; adaptive pattern recognition; competitive learning; recurrent neural networks; time-series analysis
- Abstract
The paper discusses the possibility of developing a hybrid adaptive resonance theory neural network architecture that can model the dynamics of speech perception and production starting from the sound constituents of phonemes. The architecture is composed of an adaptive resonance theory network coupled with a recurrent neural network. The hybrid network was trained to learn and generate successfully the elemental patterns of the main single vowel sounds in the English alphabet. The proposed configuration proved adequate to self-stabilize in real-time its learning independently of a teacher.
- Copyright
- © 2019, the Authors. Published by Atlantis Press.
- Open Access
- This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).
Cite this article
TY - CONF AU - Marius CRISAN PY - 2019/12 DA - 2019/12 TI - Adaptive Resonance Theory Neural Network for Phoneme Perception and Production BT - Proceedings of the 2019 2nd International Conference on Mathematics, Modeling and Simulation Technologies and Applications (MMSTA 2019) PB - Atlantis Press SP - 213 EP - 216 SN - 2352-538X UR - https://doi.org/10.2991/mmsta-19.2019.45 DO - 10.2991/mmsta-19.2019.45 ID - CRISAN2019/12 ER -