| Abstract [eng] |
The goal of this study is to separate the letters of Lithuanian words that are recorded in audio files. In other words, split audio file tracks into intervals, where the beginning and end of each interval indicate to the beginning and end of the letter. Two methods were used to do this:application of changepoints analysis for speech signal and application of changepoints analysis for fundamental frequency. First, all the letters of the words were marked manually in „Audacity“ program. The changepoints obtained using both methods were compared with the points of the manually marked letters. Comparing the signal and the fundamental frequency, it can be said that more accurate changepoints are determined in the signal. |
