Transcribers: a Free Tool for Segmenting, Labeling and Transcribing Speech


This paper describes the first version of "Transcriber", a tool for segmenting, labeling and transcribing speech. It is developed under Unix in the Tcl/Tk script language with extensions in C, and is available as free software. The environment offers the basic functions necessary for segmenting, labeling and transcribing long duration signals. The signal editor and the text editor are integrated and synchronized in order to display and play the current segment. The output is in a standard SGML format. Multiple languages are supported. The tool can be ported to various platforms and is very flexible so that new functions can be easily added. We hope that such a portable, widely available and flexible tool will benefit the whole community and make it easier to develop and share corpora.