A framework for auditory model comparability and applicability

  • Dietz M
  • Marquardt T
  • Majdak P
  • Stern RM
  • Hartmann WM
  • Dan Goodman
  • Ewert SD
Dietz M, Marquardt T, Majdak P, Stern RM, Hartmann WM, Goodman DF, Ewert SD
Acoustical Society of America (2016)
doi: 10.1121/1.4970386
Acoustical Society of America meeting 2016


Many computational models of the auditory system exist, most of which can predict a variety of psychoacoustical, physiological, or other experimental data. However, it is often challenging to apply existing third party models to own experimental paradigms, even if the model code is available. It will be demonstrated that model applicability is increased by providing a framework where the model acts as artificial observer performing exactly the same task as the subject (e.g., adaptive staircase procedure). A possible separation of the actual auditory processing of the model from the decision making stage will be discussed, which allows for testing the auditory processing of one model in a variety of experimental paradigms. The framework will consist of a citable data repository providing the required data for the models as well as toolboxes implementing both the auditory models and a variety of experimental paradigms. The model framework will be demonstrated with exemplary binaural models applied to the three most common binaural psychoacoustic paradigms: just noticeable difference (e.g., interaural time difference), tone in noise detection (e.g., binaural masking level difference), and absolute judgment (e.g., sound source localization). Further development of the framework will be discussed.


Related software

Auditory Model Initiative

Python/Matlab package for comparing binaural auditory models.