Balsių ir dvibalsių atpažinimas lietuviškoje šnekoje
Recognition of Vowels and Diphthongs in Lithuanian Speech

Summary/Abstract: Speech recognition technologies appeared in the period of general device miniaturization, when all technologies were commonly integrated into one chip. There is no space for buttons and displays anymore. In order to have a good system of speech recognition in Lithuanian a number of researches must be implemented. Only after selecting the most efficient speech recognition scheme we can proceed to the development of software adapted to the contemporary time. The aim of this paper is to determine how efficient speech recognition is possible using neuron networks. MFCC and LPC coefficients were chosen as the parameters characterizing the phonemes. The paper attempts determin the coefficients, which lead to the most efficient recognition of the phonemes. For testing programs PRAAT and MatLab were used. After the implementation of a number of phoneme recognition experiments in the research work the results were obtained, which lead to the following conclusions: 1. In case of using neuron network for the recognition of isolated sounds and characterizing the phonemes by MFCC or LPC coefficients, the possibility of recognition does not exceed 90 per cent. It is not enough for quality recognition of Lithuanian speech. 2. In case of using MFCC coefficients the phonemes are recognized better than using LPC coefficients. The average difference is about 15 percent. 3. The phoneme recognition rate increases when a speech signal is normalized before the coefficients extraction. _______: Lithuanian speech, vowels, diphthongs, speech recognition technologies, neuron networks, MFCC and LPC coefficients.

• Details
• Contents