AIDA: An Active Inference-based Design Agent for Audio Processing Algorithms

TL;DR

AIDA is an active inference-based agent that interactively personalizes audio processing algorithms, specifically tuning hearing aid parameters through Bayesian trial design and probabilistic modeling, to improve user satisfaction.

Contribution

The paper introduces a novel active inference framework with a Bayesian trial design approach for personalized audio algorithm tuning, including new generative models for acoustic signals and user responses.

Findings

Successfully implemented in a factor graph with variational message passing.

Demonstrated ability to personalize hearing aid parameters.

Accessible verification and validation experiments on GitHub.

Abstract

In this paper we present AIDA, which is an active inference-based agent that iteratively designs a personalized audio processing algorithm through situated interactions with a human client. The target application of AIDA is to propose on-the-spot the most interesting alternative values for the tuning parameters of a hearing aid (HA) algorithm, whenever a HA client is not satisfied with their HA performance. AIDA interprets searching for the "most interesting alternative" as an issue of optimal (acoustic) context-aware Bayesian trial design. In computational terms, AIDA is realized as an active inference-based agent with an Expected Free Energy criterion for trial design. This type of architecture is inspired by neuro-economic models on efficient (Bayesian) trial design in brains and implies that AIDA comprises generative probabilistic models for acoustic signals and user responses. We propose a novel generative model for acoustic signals as a sum of time-varying auto-regressive filters and a user response model based on a Gaussian Process Classifier. The full AIDA agent has been implemented in a factor graph for the generative model and all tasks (parameter learning, acoustic context classification, trial design, etc.) are realized by variational message passing on the factor graph. All verification and validation experiments and demonstrations are freely accessible at our GitHub repository.