Resynthesizing the GECO speech corpus with VocalTractLab
Lists
Only you see this
Abstract
We are addressing the challenge of learning an inverse mapping between<br> acoustic features and control parameters of a vocal tract simulator. As a first step,<br> we synthesize an articulatory corpus consisting of control parameters and wave<br> forms using VocalTractLab (VTL; [1]) as the vocal tract simulator. The basis for<br> the synthesis is a concatenative approach that combines gestures of VTL according<br> to a SAMPA transcription. SAMPA transcriptions are taken from the GECO corpus<br> [2], a spontaneous speech corpus of southern German. The presented approach uses<br> the duration of the phones and extracted pitch contours to create gesture files for the<br> VTL. The resynthesis of the GECO corpus results in 53960 valid spliced out word<br> samples totalling in 6 hours and 23 minutes of synthesized speech. The synthesis<br> quality is mediocre. We believe that the synthesized samples resemble some of the<br> natural variability found in natural human speech.
BibTeX
@inproceedings{sering2019resynthesizing,
author = {Konstantin Sering and Niels Stehwien and Yingming Gao and Martin V. Butz and R. Harald Baayen},
title = {{Resynthesizing the GECO speech corpus with VocalTractLab}},
booktitle = {Zenodo},
year = {2019},
doi = {10.5281/zenodo.4115330},
}Loading related content...