Electrical Engineering and Systems Science > Audio and Speech Processing
[Submitted on 9 Nov 2023]
Title:Sound field reconstruction using neural processes with dynamic kernels
View PDFAbstract:Accurately representing the sound field with the high spatial resolution is critical for immersive and interactive sound field reproduction technology. To minimize experimental effort, data-driven methods have been proposed to estimate sound fields from a small number of discrete observations. In particular, kernel-based methods using Gaussian Processes (GPs) with a covariance function to model spatial correlations have been used for sound field reconstruction. However, these methods have limitations due to the fixed kernels having limited expressiveness, requiring manual identification of optimal kernels for different sound fields. In this work, we propose a new approach that parameterizes GPs using a deep neural network based on Neural Processes (NPs) to reconstruct the magnitude of the sound field. This method has the advantage of dynamically learning kernels from simulated data using an attention mechanism, allowing for greater flexibility and adaptability to the acoustic properties of the sound field. Numerical experiments demonstrate that our proposed approach outperforms current methods in reconstructing accuracy, providing a promising alternative for sound field reconstruction.
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.