3ème Congrés français d'acoustique
3rd French conference on acoustics

J. Phys. IV France 04 (1994) C5-553-C5-556
DOI: 10.1051/jp4:19945119

Identification de voyelles simultanées harmoniques et inharmoniques

A. DE CHEVEIGNɹ, S. McADAMS^{2, 3}, J. LAROCHE⁴ and M. ROSENBERG<sup>4

1</sup>  Laboratoire de Linguistique Formelle, CNRS, Université Paris 7, Case 7003, 2 place Jussieu, 75251 Paris cedex 05, France
²  Laboratoire de Psychologie Expérimentale, CNRS, Université René Descartes, 28 rue Serpente, 75006 Paris, France
³  IRCAM, 1 place Stravinsky, 75004 Paris, France
⁴  Departement Signal, Télécom Paris, ENST/CNRS, 46 rue Barrault, 75634 Paris cedex 13, France

Abstract
It is known that the auditory system uses harmonicity cues to separate concurrent voiced sounds that differ in fundamental frequency (F0). What is less clear is whether the harmonicity of the target sound is used to enhance it within the mixture, or whether that of the background is used to cancel it and thus allow the target to emerge. An experiment was designed to answer this question. Subjects were presented with pairs of concurrent synthetic vowels, each of which was either harmonic or inharmonic. Results were scored according to the harmonicity of the vowel identified (the target), and that of the second vowel (the ground). For a given target, identification was best for harmonic grounds, except when target and ground were both harmonic and had the same F0. This is compatible with the cancellation hypothesis. On the other hand, identification for a given ground was worse when the target was harmonic. This is the opposite of the effect predicted by the enhancement hypothesis.

© EDP Sciences 1994