Manifold learning-supported estimation of relative transfer functions for spatial filtering
This work addresses performance degradation in RTF estimation for teleconferencing and similar audio applications, offering a method that leverages prior data, but it is incremental as it builds on existing RTF estimators with a hybrid approach.
The paper tackles the problem of estimating Relative Transfer Functions (RTFs) for spatial filtering in voice capture applications, which often degrade under adverse acoustic conditions, by using Variational Autoencoders trained on prior data from similar environments to enhance RTF estimates, achieving improved performance as confirmed by real-world experiments.
Many spatial filtering algorithms used for voice capture in, e.g., teleconferencing applications, can benefit from or even rely on knowledge of Relative Transfer Functions (RTFs). Accordingly, many RTF estimators have been proposed which, however, suffer from performance degradation under acoustically adverse conditions or need prior knowledge on the properties of the interfering sources. While state-of-the-art RTF estimators ignore prior knowledge about the acoustic enclosure, audio signal processing algorithms for teleconferencing equipment are often operating in the same or at least a similar acoustic enclosure, e.g., a car or an office, such that training data can be collected. In this contribution, we use such data to train Variational Autoencoders (VAEs) in an unsupervised manner and apply the trained VAEs to enhance imprecise RTF estimates. Furthermore, a hybrid between classic RTF estimation and the trained VAE is investigated. Comprehensive experiments with real-world data confirm the efficacy for the proposed method.