The Effects of Uncertainty in Level on Speech-on-Speech Masking

Abstract

Identification of speech from a "target" talker was measured in a speech-on-speech masking task with two simultaneous "masker" talkers. The overall level of each talker was either fixed or randomized throughout each stimulus presentation to investigate the effectiveness of level as a cue for segregating competing talkers and attending to the target. Experimental manipulations included varying the level difference between talkers and imposing three types of target level uncertainty: 1) fixed target level across trials, 2) random target level across trials, or 3) random target levels on a word-by-word basis within a trial. When the target level was predictable performance was better than corresponding conditions when the target level was uncertain. Masker confusions were consistent with a high degree of informational masking (IM). Furthermore, evidence was found for "tuning" in level and a level "release" from IM. These findings suggest that conforming to listener expectation about relative level, in addition to cues signaling talker identity, facilitates segregation of, and maintaining focus of attention on, a specific talker in multiple-talker communication situations.

Individual differences in speech-on-speech masking are correlated with cognitive and visual task performance

“…A closed-set matrix-style masked speech identification task was used ( Kidd et al , 2008a ; Byrne et al , 2022 ). The target utterances were five-word sentences using the fixed syntactic structure of name, verb, number, adjective, and object, e.g., “Sue saw two big shoes.” The speech corpus [refer to Kidd et al (2008a) ] included the natural production of eight exemplar words in each syntactic category, which had been recorded from 12 female talkers.…”

“…The use of this masker level difference rove was motivated by previous research revealing non-monotonic psychometric functions for colocated SOS tasks ( Dirks and Bower, 1969 ; Brungart, 2001 ; MacPherson and Akeroyd, 2014 ; Byrne et al , 2022 ). In all cases, performance declines as positive TMRs decrease and approach 0 dB but, in at least some listeners, can improve as the target level decreases below 0 dB and becomes noticeably quieter than the maskers (i.e., segregation by level allows the softer level signal—the target—to be intelligible).…”

“…The level difference range used here was chosen based on pilot testing and served to reduce the feasibility of attending to the lower-level stimulus (cf. Byrne et al , 2022 ).…”

“…Although group mean data clearly delineated these filter-like patterns of results, Marrone et al (2008) observed individual differences not only with respect to the maximum amount of SRM that occurred (i.e., the dB difference between the colocated reference condition and the lowest spatially separated thresholds being referred to as the range of the filter), but also with respect to the sharpness of spatial tuning itself (i.e., the bandwidths and “attenuation” characteristics of individually fit filters). Furthermore, using SRM for SOS tasks is complicated by individual differences that vary for colocated and separated stimulus presentation due in part to the different segregation cues available in the two cases [refer to Kidd and Colburn (2017) and Byrne et al (2022) ]. It currently is not known whether these finer-grained analyses of SOS masking conditions could yield an improved understanding of individual differences in performance or the extent to which cognitive factors play a role in these differences.…”

Effects of presentation level on speech-on-speech masking by voice-gender difference and spatial separation between talkers

“…In order to enhance the perception of those with auditory deficits, we must first understand the strategies utilized in these situations by normal hearing (NH) listeners. In many previous studies with NH listeners, the talkers in multi-talker situations (i.e., speech-on-speech masking situations) can be segregated by utilizing two major acoustic cues: (1) differences in voice characteristics (e.g., pitch, timbre, and loudness) among talkers ( Brungart, 2001 ; Brungart et al, 2001 ; Darwin et al, 2003 ; Mackersie et al, 2011 ; Byrne et al, 2022 ) and (2) spatial separation between target and competing talkers ( Ericson et al, 2004 ; Best et al, 2012 ; Gallun et al, 2013 ; Srinivasan et al, 2016 ; Humes et al, 2017 ). It should be noted that the aforementioned cues allow the listener a “release from masking” which refers to the ability to separate non-target speech from mixed voices and understand the target speaker.…”