PENGENALAN SUARA KE TEKS MENGGUNAKAN HIDDEN MARKOV MODEL
Total View This Week0

Institusion

Universitas Sriwijaya

Author

KURNIAWAN, IRVAN
Kurniati, Rizki
Arsalan, Osvari

Subject

P98-98.5 Computational linguistics. Natural language processing 

Datestamp

2023-04-12 02:31:19 

Abstract :
Learning new words can be difficult, it helps if it?s possible to look up the new word in a dictionary, however, a word in English often sounds alike to another word, therefore a speech to text system can help searching a word in dictionary. The use of Mel-Frequency Cepstral Coefficient in feature extraction and Hidden Markov Model in recognizing speech to text was chosen because MFCC and Hidden Markov Model has better performance compared to other speech recognition machine learning methods thus in this research, an application for text to speech was developed using the combination of MFCC and Hidden Markov Model method. There are 3 HMM model that were developed by using 3 different datasets and using same configuration. The best model acquired 100% accuracy which came from second dataset that has strong and stable voice signal, clear pronunciation, and not lot of testing data. 

File :
RAMA_55201_09021181924159.pdf
RAMA_55201_09021181924159_TURNITIN.pdf
RAMA_55201_09021181924159_0012079104_0028068806_01_front_ref.pdf
RAMA_55201_09021181924159_0012079104_0028068806_02.pdf
RAMA_55201_09021181924159_0012079104_0028068806_03.pdf
RAMA_55201_09021181924159_0012079104_0028068806_04.pdf
RAMA_55201_09021181924159_0012079104_0028068806_05.pdf
RAMA_55201_09021181924159_0012079104_0028068806_06.pdf
RAMA_55201_09021181924159_0012079104_0028068806_07_ref.pdf
RAMA_55201_09021181924159_0012079104_0028068806_08_lamp.pdf