Psychoacoustic experiments and auditory models are fundamental elements of hearing research helping to understand human auditory perception. One successful way to apply models has been to use the model as artificial observer, performing exactly the same psychoacoustic experiment as human subjects [e.g., Jepsen et al., J. Acoust. Soc. Am. 124, 422 (2008)]. While the signal processing parts of this and other models are publicly available, reproducible research requires availability of the complete framework including stimulus generation, experimental procedure, and interface to the model. For this, AFC for Matlab/Octave [www.aforcedchoice.com] provides a free and highly flexible framework to design and run psychoacoustic measurements with subjects and computer models. Previous versions of AFC have been used for nearly two decades in several highly ranked psychoacoustic research sites. To foster reproducible research, AFC offers full downward compatibility to the very first version, and the ability to easily overload or add measurement procedures, audio drivers, and models/model interfaces. Here a new version is presented with the above model as a use case. A database of psychoacoustic experiments from numerous publications is established, to provide the stimulus generation of the experiments, methods, and models for exact reproduction of the original work for teaching and research.