2019 INTERSPEECH INTERSPEECH 2019

Formant Pattern and Spectral Shape Ambiguity of Vowel Sounds, and Related Phenomena of Vowel Acoustics — Exemplary Evidence

Abstract

In the specialist literature on vowel acoustics, there is an extensive and often controversial debate on whether the primary acoustic cues of vowel quality are contained in the formant patterns or, alternatively, in the spectral shape. Yet, recent studies have shown that neither formant patterns nor spectral shapes are vowel quality-specific but that they are ambiguous because of a complex interaction between pitch and vowel-related spectral characteristics. In order to give insight into the phenomenon of formant pattern and spectral shape ambiguity of vowel sounds and its role for vowel acoustics, exemplary series of speech and of vowel sounds are presented in an online documentation, most of them selected from the Zurich Corpus. The presentation includes sound playbacks and results of an acoustic analysis (FFT spectra, LPC curves, spectrograms, fo contours, formant patterns) and of a vowel recognition test. A Klatt synthesiser is also included for resynthesis and synthesis purposes. The presentation intends (i) to support researchers in their evaluation of existing and future studies, questioning whether the actual variation and pitch-dependency of the vowel spectrum is taken into account when attempting to generalise experimental results, and (ii) to support students in their acquisition of state-of-the-art knowledge of vowel acoustics.

🌉 Interdisciplinary Bridge — Artificial Intelligence and Machine Learning and Mathematics & Optimization
🧭 Keyword Pioneer — formant pattern
🐝 Cross-Pollinator — Artificial Intelligence, Computer Science, Computer Vision, Data Science & Analytics, Deep Learning, Healthcare & Medicine, Interdisciplinary, Machine Learning, Mathematics & Optimization, Natural Language Processing, Reinforcement Learning, Robotics, Security & Privacy, Speech & Audio