{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/end-to-end-probabilistic-inference-for","title":"End-to-End Probabilistic Inference for Nonstationary Audio Analysis","arxiv_id":"1901.11436","date":"2019-01-31","proceeding":null,"authors":["William J. Wilkinson","Michael Riis Andersen","Joshua D. Reiss","Dan Stowell","Arno Solin"],"abstract":"A typical audio signal processing pipeline includes multiple disjoint\nanalysis stages, including calculation of a time-frequency representation\nfollowed by spectrogram-based feature analysis. We show how time-frequency\nanalysis and nonnegative matrix factorisation can be jointly formulated as a\nspectral mixture Gaussian process model with nonstationary priors over the\namplitude variance parameters. Further, we formulate this nonlinear model's\nstate space representation, making it amenable to infinite-horizon Gaussian\nprocess regression with approximate inference via expectation propagation,\nwhich scales linearly in the number of time steps and quadratically in the\nstate dimensionality. By doing so, we are able to process audio signals with\nhundreds of thousands of data points. We demonstrate, on various tasks with\nempirical data, how this inference scheme outperforms more standard techniques\nthat rely on extended Kalman filtering.","url_abs":"http://arxiv.org/abs/1901.11436v5","url_pdf":"http://arxiv.org/pdf/1901.11436v5.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"end-to-end-probabilistic-inference-for","repo_url":"https://github.com/AaltoML/nonstationary-audio-gp","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"audio-signal-processing","task_name":"Audio Signal Processing"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[{"method_slug":"gaussian-process","method_name":"Gaussian Process"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1901.11436","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}