Projects per year
Abstract
Modelling the spatial propagation of sound is key in applications such as personal communication, sound field reproduction, machine hearing and architecture. Especially, sound fields in enclosed spaces are of great importance, as most of our time and much of our technology is used indoors. Sound propagation in enclosed sound fields is however highly complex: Reflected, scattered and diffracted wavefronts interfere and can form complicated sound fields, which are challenging to model and analyse. This PhD thesis is concerned with the spatial analysis of measured, reverberant sound fields in enclosures. Two main objectives are pursued: Firstly, models are developed and proposed for sound field reconstruction, aiming at the interpolation and extrapolation of a reverberant sound field from a limited set of distributed microphones. The examined models are based on decomposition of the sound field in terms of overlapping subpartitions which are partially redundant. The findings indicate that sub-partition models conform to the local fine
structure in complex sound fields and allow for accurate reconstructions. The results improve further when structure is imposed on the models, such as local sparsity and continuity. It is also found that spatial functions learned in one measured sound field generalize to another sound field in a different room. Secondly, this thesis demonstrates how distributed sensor arrays can be used to analyze complex sound fields. Specifically, ad-hoc distributed microphone arrays are used to infer the location of a sound source in a room. Experiments show that local estimates from each array can be fused to accurately track the coordinates of a moving sound source inside a conventional room.
Techniques which are able to sense and exploit spatial structures of sound fields bear great potential, especially as acoustic, audio and communication technology become increasingly sophisticated. Successful analysis of the sound field surrounding a person or device can facilitate a more informed representation of space and a closer interaction with the scene, for example, in immersive, navigable, and assistive technology applications.
structure in complex sound fields and allow for accurate reconstructions. The results improve further when structure is imposed on the models, such as local sparsity and continuity. It is also found that spatial functions learned in one measured sound field generalize to another sound field in a different room. Secondly, this thesis demonstrates how distributed sensor arrays can be used to analyze complex sound fields. Specifically, ad-hoc distributed microphone arrays are used to infer the location of a sound source in a room. Experiments show that local estimates from each array can be fused to accurately track the coordinates of a moving sound source inside a conventional room.
Techniques which are able to sense and exploit spatial structures of sound fields bear great potential, especially as acoustic, audio and communication technology become increasingly sophisticated. Successful analysis of the sound field surrounding a person or device can facilitate a more informed representation of space and a closer interaction with the scene, for example, in immersive, navigable, and assistive technology applications.
Original language | English |
---|
Publisher | Technical University of Denmark |
---|---|
Number of pages | 128 |
Publication status | Published - 2022 |
Fingerprint
Dive into the research topics of 'Distributed Microphone and Array Processing in Rooms'. Together they form a unique fingerprint.Projects
- 1 Finished
-
Sound Field Analysis and Microphone Array in Processing in Rooms
Hahmann, M. (PhD Student), Fernandez Grande, E. (Main Supervisor) & Agerkvist, F. T. (Supervisor)
01/02/2019 → 03/02/2023
Project: PhD