A User-Friendly and Adaptable Re-Implementation of an Acoustic Prominence Detection and Annotation Tool

Jana Voße, Petra Wagner


Annotating prominence phenomena in speech corpora is a challenging task, as it requires many resources. Therefore, several approaches have emerged in the past decades to automatise the process of detecting and annotating prominence. Among these, [1] propose a fully automatically operating acoustic prominence detection and annotation tool that yields promising results. The present work aims at making this tool accessible to a broader community and more inviting in the manipulation of features. To do so, we re-implemented the prominence annotation approach of [1] in the programming language of the software Praat [2], which is commonly used for speech analysis purposes within several areas of research. By implementing a user-friendly interface, the Praat-based prominence detection and annotation tool can be controlled without any source code interaction, which makes it accessible to users with differing levels of programming experience. More experienced users have the option to directly work with the comprehensively commented and documented source code to manipulate or add features within the prominence detection and annotation process. Providing a more accessible and easier to manipulate re-implementation of the tool of [1], we want to contribute to further developments in the area of automatic prominence detection and annotation.


Cite as: Voße, J., Wagner, P. (2019) A User-Friendly and Adaptable Re-Implementation of an Acoustic Prominence Detection and Annotation Tool. Proc. Interspeech 2019, 3673-3674.


@inproceedings{Voße2019,
  author={Jana Voße and Petra Wagner},
  title={{A User-Friendly and Adaptable Re-Implementation of an Acoustic Prominence Detection and Annotation Tool}},
  year=2019,
  booktitle={Proc. Interspeech 2019},
  pages={3673--3674}
}