Robust multi-feature segmentation and indexing for natural sound environments

Gordon Wichern, Harvey Thornburg, Brandon Mechtley, Alex Fink, Kai Tu, Andreas Spanias

Research output: Chapter in Book/Report/Conference proceedingConference contribution

10 Citations (Scopus)

Abstract

Creating an audio database from continuous long-term recordings, allows for sounds to not only be linked by the time and place in which they were recorded, but also to sounds with similar acoustic characteristics. Of paramount importance in this application is the accurate segmentation of sound events, enabling realistic navigation of these recordings. We first propose a novel feature set of specific relevance to environmental sounds, and then develop a Bayesian framework for sound segmentation, which fuses dynamics across multiple features. This probabilistic model possesses the ability to account for non-instantaneous sound onsets and absent or delayed responses among individual features, providing flexibility in defining exactly what constitutes a sound event. Example recordings demonstrate the diversity of our feature set, and the utility of our probabilistic segmentation model in extracting sound events from both indoor and outdoor environments.

Original languageEnglish (US)
Title of host publicationCBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings
Pages69-76
Number of pages8
DOIs
StatePublished - 2007
EventCBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing - Bordeaux, France
Duration: Jun 25 2007Jun 27 2007

Other

OtherCBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing
CountryFrance
CityBordeaux
Period6/25/076/27/07

Fingerprint

Acoustic waves
Segmentation
Indexing
Electric fuses
Navigation
Acoustics
Data base
Probabilistic model

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Information Systems
  • Information Systems and Management

Cite this

Wichern, G., Thornburg, H., Mechtley, B., Fink, A., Tu, K., & Spanias, A. (2007). Robust multi-feature segmentation and indexing for natural sound environments. In CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings (pp. 69-76). [4275057] https://doi.org/10.1109/CBMI.2007.385394

Robust multi-feature segmentation and indexing for natural sound environments. / Wichern, Gordon; Thornburg, Harvey; Mechtley, Brandon; Fink, Alex; Tu, Kai; Spanias, Andreas.

CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings. 2007. p. 69-76 4275057.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Wichern, G, Thornburg, H, Mechtley, B, Fink, A, Tu, K & Spanias, A 2007, Robust multi-feature segmentation and indexing for natural sound environments. in CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings., 4275057, pp. 69-76, CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Bordeaux, France, 6/25/07. https://doi.org/10.1109/CBMI.2007.385394
Wichern G, Thornburg H, Mechtley B, Fink A, Tu K, Spanias A. Robust multi-feature segmentation and indexing for natural sound environments. In CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings. 2007. p. 69-76. 4275057 https://doi.org/10.1109/CBMI.2007.385394
Wichern, Gordon ; Thornburg, Harvey ; Mechtley, Brandon ; Fink, Alex ; Tu, Kai ; Spanias, Andreas. / Robust multi-feature segmentation and indexing for natural sound environments. CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings. 2007. pp. 69-76
@inproceedings{2e46d3c0a4374491a17d529eaae8a729,
title = "Robust multi-feature segmentation and indexing for natural sound environments",
abstract = "Creating an audio database from continuous long-term recordings, allows for sounds to not only be linked by the time and place in which they were recorded, but also to sounds with similar acoustic characteristics. Of paramount importance in this application is the accurate segmentation of sound events, enabling realistic navigation of these recordings. We first propose a novel feature set of specific relevance to environmental sounds, and then develop a Bayesian framework for sound segmentation, which fuses dynamics across multiple features. This probabilistic model possesses the ability to account for non-instantaneous sound onsets and absent or delayed responses among individual features, providing flexibility in defining exactly what constitutes a sound event. Example recordings demonstrate the diversity of our feature set, and the utility of our probabilistic segmentation model in extracting sound events from both indoor and outdoor environments.",
author = "Gordon Wichern and Harvey Thornburg and Brandon Mechtley and Alex Fink and Kai Tu and Andreas Spanias",
year = "2007",
doi = "10.1109/CBMI.2007.385394",
language = "English (US)",
isbn = "1424410118",
pages = "69--76",
booktitle = "CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings",

}

TY - GEN

T1 - Robust multi-feature segmentation and indexing for natural sound environments

AU - Wichern, Gordon

AU - Thornburg, Harvey

AU - Mechtley, Brandon

AU - Fink, Alex

AU - Tu, Kai

AU - Spanias, Andreas

PY - 2007

Y1 - 2007

N2 - Creating an audio database from continuous long-term recordings, allows for sounds to not only be linked by the time and place in which they were recorded, but also to sounds with similar acoustic characteristics. Of paramount importance in this application is the accurate segmentation of sound events, enabling realistic navigation of these recordings. We first propose a novel feature set of specific relevance to environmental sounds, and then develop a Bayesian framework for sound segmentation, which fuses dynamics across multiple features. This probabilistic model possesses the ability to account for non-instantaneous sound onsets and absent or delayed responses among individual features, providing flexibility in defining exactly what constitutes a sound event. Example recordings demonstrate the diversity of our feature set, and the utility of our probabilistic segmentation model in extracting sound events from both indoor and outdoor environments.

AB - Creating an audio database from continuous long-term recordings, allows for sounds to not only be linked by the time and place in which they were recorded, but also to sounds with similar acoustic characteristics. Of paramount importance in this application is the accurate segmentation of sound events, enabling realistic navigation of these recordings. We first propose a novel feature set of specific relevance to environmental sounds, and then develop a Bayesian framework for sound segmentation, which fuses dynamics across multiple features. This probabilistic model possesses the ability to account for non-instantaneous sound onsets and absent or delayed responses among individual features, providing flexibility in defining exactly what constitutes a sound event. Example recordings demonstrate the diversity of our feature set, and the utility of our probabilistic segmentation model in extracting sound events from both indoor and outdoor environments.

UR - http://www.scopus.com/inward/record.url?scp=46749124289&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=46749124289&partnerID=8YFLogxK

U2 - 10.1109/CBMI.2007.385394

DO - 10.1109/CBMI.2007.385394

M3 - Conference contribution

AN - SCOPUS:46749124289

SN - 1424410118

SN - 9781424410118

SP - 69

EP - 76

BT - CBMI'2007 - 2007 International Workshop on Content-Based Multimedia Indexing, Proceedings

ER -