A Probabilistic Generative Model for Spectral Speech Enhancement

Abstract

Speech enhancement in hearing aids remains a difficult task in nonstationary acoustic environments, mainly because current signal processing algorithms rely on fixed, manually tuned parameters that cannot adapt in situ to different users or listening contexts. This paper introduces a unified modular framework that formulates signal processing, learning, and personalization as Bayesian inference with explicit uncertainty tracking. The proposed framework replaces ad hoc algorithm design with a single probabilistic generative model that continuously adapts to changing acoustic conditions and user preferences. It extends spectral subtraction with principled mechanisms for in-situ personalization and adaptation to acoustic context. The system is implemented as an interconnected probabilistic state-space model, and inference is performed via variational message passing in the RxInfer.jl probabilistic programming environment, enabling real-time Bayesian processing under hearing-aid constraints. Proof-of-concept experiments on the VoiceBank+DEMAND corpus show competitive speech quality and noise reduction with 85 effective parameters. The framework provides an interpretable, data-efficient foundation for uncertainty-aware, adaptive hearing-aid processing and points toward devices that learn continuously through probabilistic inference.

Publication
IEEE Open Journal of Signal Processing, vol. 7, pp. 591–611
Marco Hidalgo Araya
Marco Hidalgo Araya
PhD student

Marco Hidalgo Araya is currently pursuing his PhD within the SPS group at Eindhoven University of Technology’s Department of Electrical Engineering.

Raphaël Trésor
Raphaël Trésor
PhD student

Raphaël Trésor is a PhD candidate in the SPS group at Eindhoven University of Technology’s Electrical Engineering department.

Bart van Erp
Bart van Erp
Chief Product Officer LazyDynamics

Bart van Erp is co-founder & product lead of Lazy Dynamics.

Wouter Nuijten
Wouter Nuijten
PhD Student

PhD student studying active inference as variational inference; core contributor to RxInfer.jl.

Thijs van de Laar
Thijs van de Laar
Assistant professor

I am an assistant professor at BIASlab, where I work on artificial agents that learn to control themselves in uncertain environments. I take inspiration from physics and neuroscience, and develop theory and (software) tools that allow for efficient, real-time interaction.

Bert de Vries
Bert de Vries
Professor

I am a professor at TU Eindhoven and team leader of BIASlab.