The Polish sentence test for speech intelligibility evaluations

Downloads

Authors

  • E. Ozimek Adam Mickiewicz University, Institute of Acoustics, Department of Room Acoustics and Psychoacoustics, Umultowska 85, 60-614 Poznań, Poland
  • D. Kutzner Adam Mickiewicz University, Institute of Acoustics, Department of Room Acoustics and Psychoacoustics, Umultowska 85, 60-614 Poznań, Poland
  • A. Sęk Adam Mickiewicz University, Institute of Acoustics, Department of Room Acoustics and Psychoacoustics, Umultowska 85, 60-614 Poznań, Poland
  • A. Wicher Adam Mickiewicz University, Institute of Acoustics, Department of Room Acoustics and Psychoacoustics, Umultowska 85, 60-614 Poznań, Poland
  • O. Szczepaniak Adam Mickiewicz University, Institute of Acoustics, Department of Room Acoustics and Psychoacoustics, Umultowska 85, 60-614 Poznań, Poland

Abstract

The main purpose of this study was to prepare Polish sentence tests for measuring speech intelligibility against an interfering noise. The tests were developed from meaningful sentences taken from everyday utterances, TV, literature etc. Two sets of sentence lists were created. The first set was optimized for the so-called binary scoring and was composed of 25 lists of 20 sentences each, while the second set was prepared for word-based scoring and was composed of 22 lists of 20 sentences each. The respective lists were statistically and phonemically balanced, i.e. they produced comparable psychometric functions and revealed comparable phonemic distribution. The mean SRT (Speech Reception Threshold) and $S_{50}$ (slope of a psychometric function at the SRT point) were: −6.1 dB and 29.5%/dB for the binary scoring and −7.4 dB and 26.7%/dB for the word-based scoring, respectively. The test lists comply with the requirements of the high quality test for measuring speech intelligibility of the Polish language.

Keywords:

speech intelligibility, sentence tests, psychometric functions, masking.