End-to-End Probabilistic Inference for Nonstationary Audio Analysis

William J. Wilkinson, Michael Riis Andersen, Joshua D. Reiss, Dan Stowell, Arno Solin

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

28 Downloads (Pure)


A typical audio signal processing pipeline includes multiple disjoint analysis stages, including calculation of a time-frequency representation followed by spectrogram-based feature analysis. We show how time-frequency analysis and nonnegative matrix factorisation can be jointly formulated as a spectral mixture Gaussian process model with nonstationary priors over the amplitude variance parameters. Further, we formulate this nonlinear model’s state space representation,
making it amenable to infinite-horizon Gaussian process regression with approximate inference via expectation propagation, which scales linearly in the number of time steps and quadratically in the state dimensionality. By doing so, we are able to process audio signals with hundreds of thousands of data points. We demonstrate, on various tasks with empirical data, how this inference scheme outperforms more standard techniques that rely on extended Kalman filtering.
Original languageEnglish
Title of host publicationProceedings of the 36th International Conference on Machine Learning
Number of pages10
Publication date2019
Publication statusPublished - 2019
Event36th International Conference on Machine Learning - Long Beach Convention Center, Long Beach, United States
Duration: 10 Jun 201915 Jun 2019
Conference number: 36


Conference36th International Conference on Machine Learning
LocationLong Beach Convention Center
CountryUnited States
CityLong Beach

Fingerprint Dive into the research topics of 'End-to-End Probabilistic Inference for Nonstationary Audio Analysis'. Together they form a unique fingerprint.

Cite this