Detailseite
Projekt Druckansicht

Identifikation früher Reflektionen mittels Sparse Recovery

Fachliche Zuordnung Akustik
Förderung Förderung von 2016 bis 2022
Projektkennung Deutsche Forschungsgemeinschaft (DFG) - Projektnummer 309256691
 
Erstellungsjahr 2021

Zusammenfassung der Projektergebnisse

The main objective of this project was to investigate into techniques for the in-situ identification of early reflections from spatial room impulse responses. Hereby, identification included the localization of early reflections and the estimation of their frequency response. We aimed at identification over the full audio bandwidth as we targeted applications as e.g. auralization, numeric simulation and psychoacoustics. The localization of early reflections was addressed using the steered-response power technique. The spatial room impulse responses are decomposed into their plane wave contributions, followed by a search for points of maximum power in this representation. Traditionally the plane wave decomposition is computed using acoustic beamforming techniques basing on linear spatiotemporal transformations. Besides these traditional techniques, we investigated the applicability of sparse recovery techniques. However, due to the underlying properties of acoustic wave propagation these techniques were not so effective for the problem at hand. As alternative, a spatially bandlimited matched-filter beamformer was proposed which shows some desirable properties for broadband direction-of-arrival estimation from spatial room impulse responses. After localization of the early reflections, their frequency dependent properties were extracted by a null-streering beamformer and fractional-octave filterbank.

Projektbezogene Publikationen (Auswahl)

 
 

Zusatzinformationen

Textvergrößerung und Kontrastanpassung