Unifying Amplitude and Phase Analysis: A Functional Multivariate Mixed-Effects Approach

• Activity Number: 494
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: IMS
• Author(s): Pantelis Hadjipantelis*+ and John Aston and Jonathan P. Evans and Hans-Georg G. Müller
• Companies: University of Warwick and University of Warwick and Academia Sinica and University of California, Davis
• Keywords: Functional Data Analysis ; Phonetic Analysis ; Multivariate Linear models

Abstract

A model for fundamental frequency (F0, or commonly pitch) within a functional data analysis framework is presented; the model accounts for both amplitude and phase variation of F0. Our approach is applied to Mandarin Chinese; a tonal language rich in pitch-related information as the relative pitch curve is specified for most syllables in the lexicon. The original five speaker corpus is preprocessed using a local smoother to produce F0 curves. These smoothed amplitude curves are then time-registered using curve synchronization techniques to remove effects of phase variation but to concurrently generate a second curve sample of curve time-registration maps. From there, both the time-registered F0 curves and their associated time-registration function curves are utilized as inputs for the computation of FPC scores and their corresponding eigenfunctions. Subsequently these scores are analyzed using a single multivariate mixed effect model that account for both amplitude and phase variations concurrently. The finding appears to confirm known tonal characteristics of the language, but interestingly suggest the presence of a sinusoid tonal component that is previously undocumented.