LESTARI, DESI
(2014)
SISTEM PENGENALAN UCAPAN HURUF VOKAL MENGGUNAKAN
METODE LINEAR PREDICTIVE CODING (LPC) DAN JARINGAN
SARAF TIRUAN LEARNING VECTOR QUANTIZATION (LVQ)
BERBASIS MIKROKONTROLER.
Other thesis, ANDALAS UNIVERSITY.
Abstract
ABSTRACT
VOWEL SPEECH RECOGNITION SYSTEM USING LINEAR
PREDICTIVE CODING (LPC) METHOD AND NEURAL NETWORK
LEARNING VECTOR QUANTIZATION (LVQ) BASED
MICROCONTROLLER
By
Desi Lestari
0910452031
Speech recognition system is the development of techniques and systems that
enable the technology to be able to accept spoken voice input, recognize and
translate. Now, speech recognition system into something that is very functional
in the field of communication technology, because speech can be a medium to
interact with the existing technology. Therefore it takes a device or system that is
able to recognize and translate the sounds of human speech.
In this final task of making speech recognition system to the sound / a / / i / / u / / e
/ and / o / by using the voice feature extraction algorithm, namely LPC. LPC is
one method of voice signal analysis stating the essential features of the voice
signal in the form of LPC coefficients. By making the process preemphasis,
windowing, autocorrelation and LPC analysis, then the obtained difference
characteristics of the speech signal coefficient values. As for the classification and
identification of speech used by the Neural Network algorithm LVQ. LVQ training
process will produce a final weight values for each vowel utterance. So the value
of the final weight will be the weight of a reference for phase identification vowel
speech recognition.
From the results of the testing that has been done to the vowel speech recognition
system is known that the introduction of a new speech utterance to lower the data
to the data of training speech. With a success rate of data speech recognition
training is 80% and for the introduction of the new pronunciation of data by 40%.
Keywords: Speech recognition, LPC, artificial neural networks, LVQ
Actions (login required)
 |
View Item |
