Transcription Guidelines for the Army Research Laboratory (ARL) SCOUT Human–Robot Dialogue Corpus

Report No. ARL-TR-8832
Authors: Claire Bonial, Cassidy Henry, Ron Artstein, Matthew Marge
Date/Pages: October 2019; 41 pages
Abstract: This report provides guidelines on how to transcribe human?robot dialogue data in the context of the US Army Combat Capabilities Development Command Army Research Laboratory Situated Corpus of Understanding Transactions (SCOUT), a developmental goal of which is to facilitate the creation of a dialogue management system onboard a robot collaborating with humans in search and navigation tasks (e.g., disaster relief). This corpus was collected via a phased "Wizard-of-Oz" (WoZ) methodology, in which human experimenters perform planned dialogue and navigation capabilities of the robot during experimental trials, unbeknownst to participants interacting with the "robot". The purpose of this report is to detail the process and guidelines in which we transcribe the speech collected for the corpus during experimentation. We describe how to use the transcription software, Praat, to complete the critical task of making the experimental speech computer-readable in text format.
Distribution: Approved for public release
  Download Report ( 2.066 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.

Last Update / Reviewed: October 1, 2019