Spherical Maps of Acoustic Properties as Feature Vectors in Machine-Learning-Based Estimation of Acoustic Parameters

Ricardo Falcon Perez*, Georg Götz, Ville Pulkki

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

38 Downloads (Pure)

Abstract

This work suggests a method of presenting information about the acoustical and geometricproperties of a room as spherical images to a machine-learning algorithm to estimate acousticalparameters of the room. The approach has the advantage that the spatial distribution of theproperties can be presented in a generic and potentially compact way to machine learningmethods. The estimation of reverberation time T60 is used as a proof-of-concept study here.The distribution of absorptive material is presented as a spherical map of feature values inwhich each value is formed by calculating the equivalent absorption area visible through thecorresponding facet of a polyhedron as seen from the polyhedron’s center point. The pixelvalues are then used as feature vectors and the real measured T60 values of correspondingrooms are used as target data. This work presents the method and trains a set of neuralnetworks with different spherical map resolutions using a dataset composed of real-worldacoustical measurements of a single room with 831 different configurations of furniture andabsorptive materials. The estimation of reverberation time using the proposed approach exhibitsa much higher accuracy compared to simple analytic methods, which proves the validity of theapproach.
Original languageEnglish
Pages (from-to)632-643
Number of pages12
JournalJournal of the Audio Engineering Society
Volume69
Issue number9
DOIs
Publication statusPublished - 6 Sep 2021
MoE publication typeA1 Journal article-refereed

Keywords

  • reverberation time
  • room acoustic modeling
  • spherical map
  • machine learning
  • room geometry

Fingerprint

Dive into the research topics of 'Spherical Maps of Acoustic Properties as Feature Vectors in Machine-Learning-Based Estimation of Acoustic Parameters'. Together they form a unique fingerprint.

Cite this