Direction of arrival estimation in reverberant environments using a single vibration sensor on an elastic panel

The vibrational response of an elastic panel to an incoming acoustic pressure wave is dependent on the coupling between the incident angle and the panel’s bending modes. By examining the relative modal excitations recorded by a single structural vibration sensor affixed to the panel, the direction o...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:The Journal of the Acoustical Society of America 2024-03, Vol.155 (3_Supplement), p.A332-A332
Hauptverfasser: Rutowski, Jenna, DiPassio, Tre, Thompson, Benjamin R., Bocko, Mark, Heilemann, Michael C.
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The vibrational response of an elastic panel to an incoming acoustic pressure wave is dependent on the coupling between the incident angle and the panel’s bending modes. By examining the relative modal excitations recorded by a single structural vibration sensor affixed to the panel, the direction of arrival (DOA) of the incident wave may be inferred. In reverberant environments, determination of the DOA of a sound source is complicated by acoustic reflections. Panel microphones may be particularly susceptible to this effect due to their large surface areas and finite modal decay times. A study on the impact of reverberation on DOA estimation using panel microphones was conducted. The panel’s response to wake word utterances was recorded in eight spaces, with reverberation times (RT60s) ranging from 0.27 to 3.00 s. These responses were used to train neural networks for DOA estimation. Results indicate an inverse relationship between RT60 and DOA estimation reliability. Within ±5°, DOA estimation reliability was measured at 95% in the least reverberant space and decreased to 78% in the most reverberant space. Results also suggest that DOA estimation using panel microphones can adapt to diverse acoustic environments by training the system with data from multiple spaces with different RT60s.
ISSN:0001-4966
1520-8524
DOI:10.1121/10.0027708