PREPOZNAVANJE POZNATIH GLASOVA STUDENATA FONETIKE U NORMALNOM GOVORU, NORMALNOM FILTRIRANOM, PRIKRIVENOM I PRIKRIVENOM FILTRIRANOM

The aim of the paper was to determine how well students of phonetics (N=J2), who know each other for more than two years, identify each others' voices in normal, normal filtered (telephone-like), disguised and disguised filtered speech. Speakers were recorded in a silent booth. They read a prepared text passage twice; once in their natural voice and once in a disguised, changed voice. A sample of 48 speech items, each 10 seconds long and acoustically different, was played randomly to each student. The identification procedure was performed individually two weeks after the recording and it was based on one listening only. The statistical significance of difference between identification scores was tested by means of Χ 2 test. The difference between acoustic parameters was determined by t-test. The results showed that the total of correct responses ranges between 65 and 90%, which meant that students of phonetics identified well. Male voices were identified more successfully across all conditions, while female listeners were more successful in their identification. Χ 2 test showed that the greatest reliable difference was between normal and disguised filtered voices (female: p<0,0001; male: p=0,0001). Male voices were equally well identified in normal and normal filtered conditions. Voices with extreme F 0 values were identified with more success than voices with average F 0 values, t-test showed that the process of acoustic filtering increased average F 0 values (p=0,03), maximum values and F 0 dispersion (male voices: p<0.01; female voices: p=0,01) as well as shimmer (p<0,01). Disguising also caused changes in acoustic parameters, more so in male voices.