Spatial release from energetic and informational masking in a divided speech identification task

Antje Ihlefeld, Barbara Shinn-Cunningham

Research output: Contribution to journalArticlepeer-review

48 Scopus citations

Abstract

When listening selectively to one talker in a two-talker environment, performance generally improves with spatial separation of the sources. The current study explores the role of spatial separation in divided listening, when listeners reported both of two simultaneous messages processed to have little spectral overlap (limiting "energetic masking" between the messages). One message was presented at a fixed level, while the other message level varied from equal to 40 dB less than that of the fixed-level message. Results demonstrate that spatial separation of the competing messages improved divided-listening performance. Most errors occurred because listeners failed to report the content of the less-intense talker. Moreover, performance generally improved as the broadband energy ratio of the variable-level to the fixed-level talker increased. The error patterns suggest that spatial separation improves the intelligibility of the less-intense talker by improving the ability to (1) hear portions of the signal that would otherwise be masked, (2) segregate the two talkers properly into separate perceptual streams, and (3) selectively focus attention on the less-intense talker. Spatial configuration did not noticeably affect the ability to report the more-intense talker, suggesting that it was processed differently than the less-intense talker, which was actively attended.

Original languageEnglish (US)
Pages (from-to)4380-4392
Number of pages13
JournalJournal of the Acoustical Society of America
Volume123
Issue number6
DOIs
StatePublished - 2008
Externally publishedYes

All Science Journal Classification (ASJC) codes

  • Arts and Humanities (miscellaneous)
  • Acoustics and Ultrasonics

Fingerprint

Dive into the research topics of 'Spatial release from energetic and informational masking in a divided speech identification task'. Together they form a unique fingerprint.

Cite this