FrAMBI: A Software Framework for Auditory Modeling Based on Bayesian Inference

Area of Science:

• Neuroscience
• Hearing Science
• Computational Auditory Neuroscience

Background:

• Auditory models are crucial for understanding listener behavior and neural mechanisms in acoustics.
• Current auditory modeling practices lack standardization, hindering reproducibility and cross-study comparisons.
• This limits the advancement of research into the neural mechanisms of sound perception.

Purpose of the Study:

• To introduce FrAMBI (Framework for Auditory Modeling based on Bayesian Inference), a novel MATLAB/Octave toolbox.
• To provide a standardized framework for implementing auditory models based on the perception-action cycle.
• To enable automatic statistical analysis of behavioral data for studying neural mechanisms of sound perception.

Main Methods:

• Development of FrAMBI, a toolbox integrated with the Auditory Modeling Toolbox (AMT).
• Implementation of auditory models using Bayesian inference within a standardized perception-action cycle structure.
• Demonstration of FrAMBI's capabilities through sound source localization tasks with varying complexity.

Main Results:

• FrAMBI successfully facilitates the implementation of auditory models for static and dynamic acoustic scenarios.
• The framework supports the definition and comparison of multiple model variants to test different neural mechanisms.
• Parameter estimation and model comparison procedures are integrated within the FrAMBI toolbox.

Conclusions:

• FrAMBI offers a standardized approach to auditory modeling, promoting reproducible research in neuroscience.
• The toolbox enhances the ability to study neural mechanisms underlying sound perception and listener behavior.
• Long-term maintenance and expansion within AMT will foster continued advancements in the field.