Acoustic Environment Classification

School of Computing Sciences, University of East Anglia

UEA logo

Series 1
- Bar
- Beach
- Bus
- Car
- Football Match
- Laundrette
- Lecture
- Office
- Rail station
- Street
Series 2
- Building site
- Bus
- Car (city)
- Car (highway)
- Launderette
- Office
- Presentation
- Shopping mall
- Street (people)
- Street (traffic)
- Supermarket
- Train

Overview of the project

Acoustic environments provide many valuable cues for context aware computing applications. From the acoustic environment we can infer the types of activity, communication modes and other actors involved in the activity. The datasets we collected are all available here.
Our initial experiments were conducted with high quality mono sound recordings made in the Norwich area (with the exception of some of the football match recordings) during the spring and summer of 2002.
For the second, low bandwidth, series of experiments we used an MP3 recorder attached to the strap of a shoulder bag as the recording device to capture the environmental noise from a typical daily routine. The recordings were made in and around Norwich area during the spring and summer of 2004. In each environment we conducted several recording sessions to gather a range of data at different times in similar places.
Environmental or background noise can be classified with a high degree of accuracy using recordings from microphones commonly found in PDAs and other consumer devices.

Environmental noise data sets

Series 1 was recorded using a Sony MiniDisk recorder and external microphone in 2002.
Sampling rate: WAV 22.050kHZ 16bit Mono
Series 2 was taken using a Samsung YP55H MP3 recorder in 2004.
Sampling rate: WAV 8.00kHZ 8bit Mono

Publications

Dan Smith, Ling Ma, Nick Ryan, Acoustic environment as an indicator of social and physical context, Personal and Ubiquitous Computing, 10(1), 2005 (DOI: 10.1007/s00779-005-0045-4)
J. Steward, Using a PDA as an Audio CaptureDevice, BSc Final Project, UEA School of Computing Sciences, 2005 (PDF 486K)
Ling Ma, Dan Smith and Ben Milner. Context Awareness using Environmental Noise Classification, Proc. Eurospeech 2003, Geneva, Switzerland, 2237-2240, 2003 (PDF 376K)
Ling Ma, Dan Smith and Ben Milner. Environmental Noise Classification for Context-Aware Applications, Proc. DEXA 2003, (LNCS 2736), 360-370, 2003 (PDF 256K)

Contact us

Dr. Dan Smith
Dr. Ben Milner
Ling Ma

Links

RWCP sound scene databse
HTK
TRECVid home page