ASA 129th Meeting - Washington, DC - 1995 May 30 .. Jun 06

5aSC5. Three acoustically predictable factors underlying vowel tongue shapes.

David A. Nix

George J. Papcun

Los Alamos Natl. Lab., CIC-3, MS B265, Los Alamos, NM 87545

To obtain a low-dimensional, speaker-independent parameterization of vowel tongue shapes, the three-mode factor analysis procedure PARAFAC [Harshman et al., 693--707 (1977)] was applied to x-ray microbeam tongue measurements of ten English vowels spoken by two male and two female subjects in seven different /CVC/ contexts. PARAFAC reliably extracts three speaker-independent, nonorthogonal factors. The resulting speaker-independent factor coefficients cluster by vowel in three-dimensional articulatory space. In two-dimensional projections, they qualitatively reflect the traditional vowel quality chart. A multi-layer perception (neural network) independently corroborates this solution: these tongue shape coefficients are significantly more predictable from the corresponding acoustics than coefficients from either the nonorthogonal two-factor solution or the orthogonally constrained three-factor solution. These three factors also correspond qualitatively to the three nonorthogonal factors extracted from Icelandic x-ray film vowel data [Jackson, 124--143 (1988)]. Thus the current English solution contradicts Jackson's distinction (based on Harshman et al.'s two-factor English vowel solution) between language-independent and language-specific vowel articulation primes. This low-dimensional, potentially cross-linguistic representation could benefit speech recognition, coding, or synthesis applications in which an acoustically correlated vowel tongue shape parameterization is required.