Unified Praat Script Tools for Facilitating Korean ToBI Annotation
Kyuchul Yoon · Seoul National University the Journal of Humanites · 2006
The purpose of this paper is to introduce a set of Praat [1] scripts developed to facilitate a large scale speech corpus annotation using the Korean ToBI prosody labeling conventions [2]. The scripts were intended to help labelers by expediting some or all of the steps required in the labeling process. The scripts are “unified” in the sense that, once the text part of the corpus is romanized, all of the corpus-building process can proceed with the scripts introduced in this work. This process may include a range of procedures from tokenizing the text paragraph and its associated long sound file into sentence-size chunks to evaluating inter-labeler agreement [3]. The scripts are open-source in nature and the users are encouraged to modify them to meet their respective needs under the terms of GNU General Public License (GPL) [10]. *