DocumentCode
727925
Title
Different strategies in the development of ANFIS to recognize vowels
Author
Valencia-Jimenez, Jesus Julian ; Fernandez-Caballero, Antonio
Author_Institution
Inst. de Investig. en Inf. de Albacete, Univ. de Castilla-La Mancha, Albacete, Spain
fYear
2015
fDate
17-20 June 2015
Firstpage
1
Lastpage
6
Abstract
This document is the result of a first approach to recognizing vowels through using Adaptive Neuro-Fuzzy Inference Systems (ANFIS), applying different strategies to train these systems. The tests perform with data from a same speaker used for training and checking; also from another speaker with different gender for checking. Training data are established minimally. Different strategies are applied and the use of cloning with variation of data is verified. The results provide a comparison of different strategies to improve vowel recognition on a specific speaker who trained the system, and another speaker of different gender who did not participate during the training of the system.
Keywords
fuzzy neural nets; fuzzy reasoning; gender issues; natural language processing; speaker recognition; ANFIS development; adaptive neuro-fuzzy inference systems; speaker gender; speaker recognition; vowel recognition; Adaptive systems; Artificial neural networks; Hidden Markov models; MATLAB; Mathematical model; Silicon compounds; Training; ANFIS; neuro-fuzzy systems; speaker recognition; speech processing; speech recognition;
fLanguage
English
Publisher
ieee
Conference_Titel
Information Systems and Technologies (CISTI), 2015 10th Iberian Conference on
Conference_Location
Aveiro
Type
conf
DOI
10.1109/CISTI.2015.7170607
Filename
7170607
Link To Document