Correlation of Speech Intelligibility Tests in Reverberant Rooms with Three Predictive Algorithms

Jacob KD. Correlation of Speech Intelligibility Tests in Reverberant Rooms with Three Predictive Algorithms. J Audio Eng Soc. 1989;37(12):1020-1030. Available from: https://aes.org/publications/elibrary-page/?id=6056

@article{Jacob1989_6056,
  author = {Jacob, Kenneth D.},
  title = {{Correlation of Speech Intelligibility Tests in Reverberant Rooms with Three Predictive Algorithms}},
  journal = {Journal of the Audio Engineering Society},
  volume = {37},
  number = {12},
  pages = {1020--1030},
  year = {1989},
  month = dec,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=6056}
}

TY  - JOUR
TI  - Correlation of Speech Intelligibility Tests in Reverberant Rooms with Three Predictive Algorithms
AU  - Jacob, Kenneth D.
T2  - Journal of the Audio Engineering Society
J2  - J. Audio Eng. Soc.
VL  - 37
IS  - 12
SP  - 1020
EP  - 1030
PY  - 1989
DA  - 1989/12/06
UR  - https://aes.org/publications/elibrary-page/?id=6056
PB  - Audio Engineering Society
LA  - en
AB  - The data base used in an earlier speech intelligibility study has been extended. The rooms added in the present study involve higher reverberation times and represent more difficult environments for sound reinforcement. The results of subjective intelligibility tests are compared to the predictions of three published algorithms. One of these predictors, based on the modulation transfer function, is being considered as an international standard (IEC Report 268-16, 1988), and its accuracy is confirmed in this study. A second predictor, based on the Lochner-Burger signal-to-noise ratio, is shown to be equally accurate when data are analyzed in a manner similar to that of two other intelligibility studies. The third algorithm, which assumes that all reflections degrade intelligibility, is shown to be inferior. The study confirms the beneficial effect of early reflections and the detrimental effect of late reflections on speech intelligibility assumed by the two more accurate algorithms.
ER  -