Effect Of Using Window Type On Time Scale Modification On Voice Recording Using Waveform Similarity Overlap and Add

Every people has different ability to listen and pronounce speech. There is a pronunciation of speech quickly or slowly. As well as hearing, some people can hear normally and some of the hearing has decreased due to heredity, age, illness and so on. In this research, time stretching process on the sound recording will be performed, which is the time noise signal density shift without changing the basic frequency using Waveform Similarity Overlap and Add (WSOLA) method. The determination to use a good window type is used for certain conditions to avoid discontinuity between frames and vibrating sounds. Mean Opinion Score (MOS) shows that the proposed hamming and triangular is superior to the rectangular at 25% −75% spacing conditions, while at 100% frame spacing, rectangular windows are superior to hamming and triangular windows.