IU_ULTRASOUND data set

Introduction

Welcome to the website of the IU_ULTRASOUND data set. This is a corpus of articulatory data recorded by 2D ultrasound during speech production, acquired from one male American English (SML) and one male Hungarian speaker (CSTG). The data was recorded at the Speech Production Lab of Indiana University (IU), Bloomington, USA, between May - July 2014.


Purpose

The primary purpose of this website is to provide data for examining the relationship of tongue shape and position with the phonetic properties of speech.


Contents

  • Ultrasound recording as JPG files
  • Speech recording as WAV file
  • EGG recording as WAV file (only for speaker CSTG)
  • Synchronized speech and ultrasound video as AVI file
  • Text prompt as TXT file

Sample

First sentence of speaker CSTG as zip file


Developers

  • Steven M. Lulich (IU)
  • Tamás Gábor Csapó (BME)

Get access

In order to get access to the IU_ULTRASOUND data, please contact Tamás Gábor Csapó.


Acknowledgements

During recording the data set, Tamás Gábor Csapó was supported by a Fulbright scholarship.


Related publications

·   Tamás Gábor Csapó, Dávid Csopor, ,,Ultrahangos nyelvkontúr követés automatikusan: a mély neuronhálókon alapuló AutoTrace eljárás vizsgálata'' (Automatic tongue contour tracking based on ultrasound: investigation of the Deep Neural Network based AutoTrace method - in Hungarian), Beszédkutatás 2015 (Speech Research 2015), Budapest, 2015, accepted.