Content-Based Classification, Search, and Retrieval of Audio

  • Authors:
  • Erling Wold;Thom Blum;Douglas Keislar;James Wheaton

  • Affiliations:
  • -;-;-;-

  • Venue:
  • IEEE MultiMedia
  • Year:
  • 1996

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many audio and multimedia applications would benefit from the ability to classify and search for audio based on characteristics of the audio rather than by keywords. These include multimedia databases and file systems, digital libraries, automatic segmentation or indexing of video (for example, news or sports footage) using the audio soundtrack, surveillance, and sound browsers for effects designers and musicians. This article describes an audio analysis, search, and classification engine that reduces sounds to perceptual and acoustical features. Sounds can then be searched or retrieved by any one or a combination of the features, by specifying previously learned classes based on these features, or by selecting or entering reference sounds and asking the engine to retrieve sounds that are similar (or dissimilar) to them. We present examples of this engine as it would be used in some of the application areas listed above. Readers may contact Erling Wold at Muscle Fish LLC, 2550 Ninth Street, Suite 207B, Berkeley, CA 94710, e-mail erling@musclefish.com.