While long-term fundamental frequency statistics have been shown to be useful in discriminating speakers, relatively little attention has been paid in work on speaker characterization to intonation as a linguistically and phonetically structured phenomenon. To help redress the balance, this article presents the results of an experiment on between-speaker differences in linguistically specified intonational events. An autosegmental-metrical model of intonation is assumed which uses H (high) and L (low) targets as its primes. Since the pitch of the events corresponding to these targets is highly variable due to factors such as within-speaker variation in pitch range (or ‘pitch span’), this study investigates as a source of speaker discrimination the temporal alignment of these intonational events with segmental events. A limited degree of discrimination is achieved in highly controlled materials. Of theoretical interest is that definable pitch events lying between H and L targets show more potential for betweenspeaker discrimination than the targets themselves.
[1]
J. Wolf.
Efficient Acoustic Parameters for Speaker Recognition
,
1972
.
[2]
J. Kraayeveld,et al.
Idiosyncrasy in prosody: speaker and speaker group identification in Dutch using melodic and temporal information
,
1997
.
[3]
Hermann J. Künzel,et al.
Sprechererkennung : Grundzüge forensischer Sprachverarbeitung
,
1987
.
[4]
Harry Hollien,et al.
The Acoustics of Crime: The New Science of Forensic Phonetics
,
1990
.
[5]
B. Atal.
Automatic Speaker Recognition Based on Pitch Contours
,
1969
.
[6]
松坂 ヒロシ.
J. D. O'Connor and G. F. Arnold : Intonation of Colloquial English, 2nd ed.
,
1977
.
[7]
J R Baldwin.
Phonetics and Speaker Identification
,
1979,
Medicine, science, and the law.