Repository logo

Infoscience

  • English
  • French
Log In
Logo EPFL, École polytechnique fédérale de Lausanne

Infoscience

  • English
  • French
Log In
  1. Home
  2. Academic and Research Output
  3. Reports, Documentation, and Standards
  4. Observations on Multi-Band Asynchrony in Distant Speech Recordings
 
report

Observations on Multi-Band Asynchrony in Distant Speech Recordings

Lathoud, Guillaume  
2006

Whenever the speech signal is captured by a microphone distant from the user, the acoustic response of the room introduces significant distortions. To remove these distortions from the signal, solutions exist that greatly improve the ASR performance (what was said?), such as dereverberation or beamforming. It may seem natural to apply those signal-level methods in the context of speaker clustering (who spoke when?) with distant microphones, for example when annotating a meeting recording for enhanced browsing experience. Unfortunately, on a corpus of real meeting recordings, it appeared that neither dereverberation nor beamforming gave any improvement on the speaker clustering task. The present technical report constitutes a first attempt to explain this failure, through a cross-correlation analysis between close-talking and distant microphone signals. The various frequency bands of the speech spectrum appear to become desynchronized when the speaker is 1 or 2 meters away from the microphone. Further directions of research are suggested to model this desynchronization.

  • Files
  • Details
  • Metrics
Loading...
Thumbnail Image
Name

rr-06-74.pdf

Access type

openaccess

Size

670.69 KB

Format

Adobe PDF

Checksum (MD5)

05988d46867dbae4bd52e7ebaca01007

Logo EPFL, École polytechnique fédérale de Lausanne
  • Contact
  • infoscience@epfl.ch

  • Follow us on Facebook
  • Follow us on Instagram
  • Follow us on LinkedIn
  • Follow us on X
  • Follow us on Youtube
AccessibilityLegal noticePrivacy policyCookie settingsEnd User AgreementGet helpFeedback

Infoscience is a service managed and provided by the Library and IT Services of EPFL. © EPFL, tous droits réservés