Thesis Archive
Filipino to Chinise Speech-to-Speech Translator Using NeuralNetwork and Database System (2012)
ESG-08-1112-01
Mark Renier M. Bailon, Mark Albert C. De Silva, Reyann Jhorel P. Lapuz, John Larry A. Tinio and Ted Bryan K. Yu, advised by Engr. Reggie C. Gustilo
Abstract:
-Technology nowadays becomes convenient because conveying information between computers and other electronic devices can be done in a speech-to-speech environment through speech implementation. There aredifferent technologies that can aid in speech implementation and one of which is the Neural NetworkSystem (NNS). A Neural Network System is composed of many simple processors combined together thatrecognizes pattern through series of trainings.The aim of the research is to produce computer software that uses an algorithm in order to translate Filipino speech into Chinese speech with the use of an artificialneural network aided by LabVIEW. The use of LabVIEWin implementing the whole speech-to-speech system hasbeen observed in detail through three stages: speech-totextstage, text-to-text stage, and text-to-speech stage. Theresults showed that increasing the number of training ofwords increases the success rate of the system inrecognizing the words. Some words which are nearlysimilar in pronunciation received lower success rates ascompared to the other words. Training has a success rateof 93.4% while testing of the 500 words was done to 25samples and it garnered an average of 91.59 percent.Speech implementation has been successfully conductedthat applies speech recognition, speech synthesis, anddatabase system. The results of the speech-to-speechtranslation of Neural Network System in LabVIEW arepresented such that they can be used as an aid toimprovement of speech implementation.
Correspondence:
Engr. Reggie C. Gustilo
[email protected]