Spatial perception of speech in various signal to noise ratios

Kim S. Abouchacra, Diana C. Emanuel, Ingrid M. Blood, Tomasz R. Letowski

Research output: Contribution to journalArticlepeer-review

17 Scopus citations


Objective: The study was designed to assess the effects of background noise level on the detection and localization of speech. Design: The phrase 'Where is this?' was presented either in quiet or in a diffuse noise field, through loudspeakers arranged in a 360°azimuth array. The noise conditions included 11 signal to noise ratios (SNRs) ranging from -18 dB SNR to +12 dB SNR in 3 dB increments. Seventeen normal-hearing subjects, aged 18 to 29, participated in the study. Results: Results revealed that in all listening conditions the signal was most easily detected when presented through a loudspeaker positioned at 90°or 270°azimuth. Although the actual level for 50% detection varied as a function of loudspeaker location and SNR, 85% and 100% of all presentations of the signal were detected at -9 dB and -6 dB SNR, respectively. Localization accuracy improved as the SNR increased, ranging from 18% accuracy at -18 dB SNR to 89% at +12 dB SNR. Localization accuracy in quiet was 95%. The data are discussed in reference to patterns of responses at each loudspeaker location. Conclusions: Detection of the target signal deteriorated as background noise level increased and was dependent on the source location of the incoming signal, as expected. Localization accuracy of the target signal was highly dependent on the SNR and spatial location of the signal source. Detection and localization accuracy data were found to be repeatable across test sessions and response patterns were found to be symmetrical on the right and left sides of the horizontal plane.

Original languageEnglish (US)
Pages (from-to)298-309
Number of pages12
JournalEar and Hearing
Issue number4
StatePublished - Aug 1998

All Science Journal Classification (ASJC) codes

  • Otorhinolaryngology
  • Speech and Hearing


Dive into the research topics of 'Spatial perception of speech in various signal to noise ratios'. Together they form a unique fingerprint.

Cite this