Interspeech'2005 - Eurospeech

Lisbon, Portugal
September 4-8, 2005

Modeling of Between-Speaker and Within-Speaker Variation in Spontaneous Speech Tempo

Hugo Quene

Utrecht University, The Netherlands

Speech tempo (speaking rate) varies both between and within speakers. Previous research suggests several relevant factors and predictors. The present study investigates all these factors combined, both between and within speakers, in a large corpus of spoken Dutch interviews. This is done by means of multi-level modeling of sex, age, and dialect region (all between speakers) and phrase length and sequential position of phrase within session (both within speakers). Results show that speech tempo depends mainly on phrase length, and not on between-speaker factors sex, age, or dialect region. Within-speaker tempo variations exceed the JND. Separate modeling of phrase length itself reveals significant negative effects of age and of sequential position, but not of region or sex. When taken together, these results underline the phonetic and communicative importance of within-speaker variations in speech tempo.

Full Paper

Bibliographic reference.  Quene, Hugo (2005): "Modeling of between-speaker and within-speaker variation in spontaneous speech tempo", In INTERSPEECH-2005, 2457-2460.