Applied Computational Intelligence and Soft Computing / 2012 / Article / Alg 1

Research Article

Segmentation and Classification of Vowel Phonemes of Assamese Speech Using a Hybrid Neural Framework

Algorithm 1

SOM weight vector extraction algorithm.
(1) Input: Spoken word of size , sampling
    frequency , duration second
(2) Preprocess the signal using preprocessing
  algorithms described in Section 7.1
(3) Initialize , order of linear prediction
(4) Find the coefficients of a th-order linear predictor FIR filter,
   that predicts the current value of the real-valued
   preprocessed time series , based on past samples.
(5) Store
(6) Take an topology map whose neurons are arranged in
  an -dimensional hexagonal pattern.
(7) Initialize weight to a small random number.
(8) Initialize learning parameter, and neighbors,
(9) for to ( ) do
    pick a(k)
  find Wining neuron as,
 Update synaptic vectors of the wining cluster,
        Update ,
(10) Store updated weight as , where