Reducing informational masking by sound segregation

Abstract

Informational masking was reduced using three stimulus presentation schemes that were intended to perceptually segregate the signal from the masker. The maskers were sets of sinusoids chosen randomly in frequency and intensity on each stimulus interval or, in some conditions, on every masker burst in a series of bursts within intervals. Masker components were excluded from the frequency region surrounding the 1000-Hz signal to minimize the energetic masking. Masked thresholds as great as 60–70 dB above quiet threshold were observed for some subjects in some conditions. It was shown that this informational masking could be reduced as much as 40 dB by: (1) presenting the masker to both ears and signal to one ear; (2) playing different masker samples sequentially in each interval of every trial; or (3) presenting the signal in alternate bursts of multiple, identical masker samples. For the binaural manipulation, informational masking was reduced because the masker and signal were perceived as originating from different interaural locations. In the latter two manipulations, a difference in the spectral or temporal pattern of the signal and masker provided the detection cue. These effects were interpreted as evidence of the importance of perceptual segregation of sounds in noisy listening environments where signal reception is not limited by energetic masking.

Temporal coherence versus harmonicity in auditory stream formation

“…Lastly, and consistent with previous findings (Kidd et al, 1994;Kidd et al, 2003;Turgeon et al, 2005), the introduction of (40-ms) onset asynchronies between the target and masker tones had a beneficial effect on sensitivity for all conditions [2.67 < t(7) < 9.87, p 0:032], including conditions with a mistuned target and harmonic maskers [M-Sync vs M-Async: t(7) ¼ 2.67, p ¼ 0.032]. This is seen by comparing the rightmost group of bars with the bars for corresponding synchronous conditions, on the left.…”

“…A similar effect was observed for the L-Sync condition, which is also consistent with the TCT , and in line with earlier findings using inharmonic MBD and MBS stimuli (Kidd et al, 1994;Kidd et al, 2003). For the frequency-shifted (S-Sync)-and thus, inharmonic-masker condition, a trend in the same direction was visible, but the difference was not statistically significant [t(7) ¼ 1.97, p ¼ 0.090]; therefore, the results of this condition are inconclusive.…”

“…Depending on the condition being tested, N was constant within each trial (Mm. 2) (conditions denoted "MBS" for "multiple bursts same," following Kidd et al, 1994;Kidd et al, 2003), or was varied randomly across bursts within trial (Mm. 3) (conditions denoted "MBD" for "multiple bursts different"), with the constraint that two consecutive N values could not be the same, with the exception of the penultimate and final burst, for which N was always the same.…”

“…Depending on the condition tested, the background tones were either temporally coherent or incoherent with the targets, and they were either harmonically related to the targets or not-with the exception of the final burst in each sequence, for which the frequencies of the masker tones were always randomly jittered. In general, repeating target tones can be "heard out" from a multi-tone background when the targets form a separate stream (e.g., Kidd et al, 1994;Kidd et al, 2003;Micheyl et al, 2007). We therefore reasoned that listeners would show good performance only in conditions in which they were able to hear the target tones as a separate stream.…”

How Visual Cues for when to Listen Aid Selective Auditory Attention

“…Only when the masker was perceived as coming from a different direction than the target did the visual cue fail to provide any significant benefit to target identification. Perceived spatial separation can yield large benefits for listeners in complex settings; it improves the ability to understand and recognize a target in the presence of an otherwise confusable masker (Kidd et al 1994;Freyman et al 1999;Best et al 2005), presumably by helping to segregate the target from the masker and allowing listeners to focus selective attention on the target (Shinn-Cunningham 2008). The current results might suggest that visual cues for when to listen provide benefits that are similar to spatial cues, helping listeners to focus on the song syllables that correspond to the target song and segregate target from masker.…”