Spatially separated sources reveal speech components audible only via binaural processing

Evidence

An experiment in which spectro-temporal regions from the target speech signal were removed according to a variable local SNR criterion gave no indication for any BMLD-like detection advantage. The results of the experiment showed a decrease in speech intelligibility when elements with a local SNR of 0 dB and larger were removed from the target speech, suggesting that speech elements at local SNRs above ‑4 dB did contribute to speech intelligibility. Interestingly, this was the case for both the collocated and spatially separated speaker configurations. The spatial release from masking appeared to be independent of the extent to which lower-SNR components were removed from the target signal. — Dijk, Pim van, Physiology, Psychoacoustics and…, 1619 (#corroborates, via linked claim)

Categorization

• corroborated: 2 independent passages support this claim
• XREF: connects to: Spatial location cues alone can segregate identical sounds into separate perceived sources — Dijk, Pim van, Physiology, Psychoacoustics and…, 4883; Binaural detection advantages do not drive the spatial speech benefit in multitalker settings; Binaural listening improves source audibility by 0–8 dB in multi-source scenes (+1 more)

Related

• Spatial release from masking persists even when target speech components are selectively discarded (similarity 0.92)
• Spatially separated speech sources improve intelligibility through binaural release from masking. (similarity 0.91)
• Binaural detection advantages do not drive the spatial speech benefit in multitalker settings (similarity 0.91)
• Directional source allocation, not binaural unmasking, explains spatial release (similarity 0.88)
• Binaural unmasking cannot explain spatial release from masking in multispeaker situations (similarity 0.87)
• Discarding low-SNR speech components affects binaural intelligibility (similarity 0.87)
• Speech elements near masked threshold carry intelligibility only under binaural listening. (similarity 0.86)
• Spatial separation of talkers reduces ambiguity in assigning speech components. (similarity 0.86)
• Spatial separation reduces inter-source confusion rather than restoring lost signal (similarity 0.85)
• Binaural listening improves source audibility by 0–8 dB in multi-source scenes (similarity 0.85)
• Spatial release of masking survives when only loud, monaurally detectable speech remains. (similarity 0.84)
• Spatial release from masking stays constant across all SNR criteria tested (similarity 0.84)
• Spatial location cues alone can segregate identical sounds into separate perceived sources (similarity 0.84)
• Binaural unmasking depends on masker duration but not signal duration (similarity 0.84)
• Binaural listening yields a 5-7 dB intelligibility advantage over monaural (similarity 0.84)

Broader Connections

• Spatial release from masking persists even when target speech components are selectively discarded #corroborates (0.80) — Dijk, Pim van, Physiology, Psychoacoustics and…, 1619
• Binaural listening improves source audibility by 0–8 dB in multi-source scenes #expands (0.75)