Cross-linguistic Speaker Identification Based On Monophthongal Vowels

Humans can identify a particular speaker with considerable ease by fairly short speech segments,but tell with great difficulty which voice information cues the speaker identity.This study investigates if reliable identification of a speaker can be achieved solely by an individual vowel.In this research,ten monophthongs are selected as target vowels for an identification test,and these vowel stimuli are recorded in a/hVd/word frame.The American,Dutch,and Chinese speakers are identified after each stimulus by native Chinese-speaking listeners.The results of the identification tests show that in a cross-linguistic setting where both native and non-native speakers of English are identified with English,vowels facilitate the identification.The acoustic analysis also indicates that the higher-frequency formants demonstrate more inter-speaker variability than the lower-frequency formants.Further,the height feature of vowels is more important than the backness feature for cross-linguistic speaker identification.Results also show that in comparison with non-Chinese speakers,the Chinese speakers are better identified by Chinese listeners who share the same native language with the speaker.