INTERSPEECH 2019
Formant pattern and spectral shape ambiguity of vowel sounds
and related phenomena of vowel acoustics
Exemplary evidence
Dieter Maurer (1), Heidy Suter (1), Christian d'Heureuse (1), Volker Dellwo (2)
(1) Institute for the Performing Arts and Film, Zurich University of the Arts, Switzerland
(2) Department of Computational Linguistics, University of Zurich, Switzerland
In contrast to the overwhelming majority of models concerning the acoustic representation of vowel
quality, single formant patterns and spectral shapes do not relate to single but to different
vowel qualities — they are ambiguous.
First and foremost, the ambiguity phenomenon is a consequence of an inherent pitch-dependency of
vowel-related spectral characteristics. This is true for inter- as well as intra-speaker comparisons
of vowel sounds and for natural as well as synthesised sounds.
The present online documentation gives insight into the phenomenon: It allows to acquire basic
knowledge, to extensively listen to exemplary sound series, to directly compare the related
formant patterns and spectral shapes, and to crosscheck the perceptual role of fundamental
frequency and formant patterns in vowel synthesis (Klatt synthesis).