volume 123, issue 5, P3877-3877 2008
DOI: 10.1121/1.2935781
View full text

Abstract: Speaker-dependent parameters (F3-F4) are expected to show a lower SD when grouped by speaker/session, Vowel quality parameters (F1-F2) are supposed to have a lower SD when grouped by phoneme. F0 would be related both to the speaker and the vowel quality. Results show that F2 is the parameter with the highest SD. F4 the one with lowest SD. F0 is highly variable between vowels. No significant differentes are found in any of the parameters when grouping by session or by speaker. Speaker/session clustering (all vo…

expand abstract