Improvement in Speech to Text for Bahasa Indonesia Through Homophone Impairment Training

Intan Sari Areni ； Indrabayu ； Anugrayani Bustamin

《電腦學刊》 28卷5期 (2017/10) Pp. 1-10

https://doi.org/10.3966/199115992017102805001

摘要

In this research, an approach for increasing accuracy in speech to text application is done using Mel Frequency Cepstral Coefficient (MFCC) trained by Backpropagation Neural Network (BPNN). A set of Bahasa Indonesia homophones data speech is used for training and validation. The record is taken from 6 native adults comprising 3 males and 3 females. Working in 16 KHz sampling mode, the data is stored in WAV format. A confusion matrix is used to validate the system with and without homophone locking learning. A significant improvement is observed from the experiment. The percentage of accuracy is increased from 53.33 to 93.4 from male samples. From females’ records, the increment is even higher. The accuracy percentage has risen from 36.8 to 93.33.

關鍵字

BPNN ； confusion matrix ； homophone ； MFCC ； speech to text