4aPP17. Effect of signal frequency uncertainty for random multi-burst maskers Rong Huang and Virginia M. Richards

1. The detectability of a coherent target of sequential tone pips amongst randomly presented tone pips (masker) is estimated. This task may reflect listeners’ ability to segregate a single sound source from competing sources that are random in time and frequency. p 0.7 0.6 0.5 0.4 0.3 0 0.48 0.79 1.08 2.31 1 0.27 0.74 0.69 1.65 2.02 delay (ms) 2 0.57 0.92 1.66 2.43 3 0.99 1.79 1.82 2.59 4 1.79 2.88 3.10 3.37 5 3.21 3.16 3.10 2.23 4aPP17. Effect of signal frequency uncertainty for random multi-burst maskers Rong Huang and Virginia M. Richards Department of Psychology, University of Pennsylvania Introduction: Experiment III: Joint effect of changes in p and in target delay Experiment I: Target Frequency Uncertainty Thresholds of p are measured for 5 known target frequencies and a Ran condition in which the target frequency is chosen at random. Four normal-hearing listeners participated. The (known) target frequency is 1000 Hz. d’ values are estimated for many p / target delay pairs. 0.2 masker alone target + masker 0.3 5000 0.4 Threshold p(d’ = 1) 0.5 Frequency (Hz) 0.6 200 0.7 445 220 1000 2245 4490 0 240 0 240 Ran MidHigh Low MidLow Mid High no target coherent target Time (ms) Target Frequency (Hz) Figure 1: Stimuli for two intervals of a trial Table I: Average (N=4) d’ values for joint changes in p and delay. Figure 3: Thresholdp as a function of frequency and frequency uncertainty The independent variable is the probability (p) that a tone pip falls in any time-by-frequency cell. As p increases, the task becomes more difficult (e.g., Kidd et al., 1995). Experiment II: Target delay relative to masker • Results and Conclusions • Thresholds for the detection of a coherent sequence of fixed- • frequency tone pips are lower for known target frequencies vs. • randomly chosen frequencies. (Figure 3) • Thresholds are lowest for the 1000-Hz target, which is the central • frequency in the current experiment (Figure 3). • Except for the lowest frequency, very small target delay • (1 ms or so) increases sensitivity from d’=0.5 to d’=1 (Figure 5). • This result is notable because the delay is between • (a) a single target tone pip & relatively few masker components • (b) the maskers have randomly chosen frequencies across time • Sensitivity relies on delay and p in an orderly way – these two • parameters trade with one another (Table I). • Citations: • Kidd,G., Mason,C.R., and Dai, H (1995). Discriminating Coherence in Spectro-Temporal Patterns, J. Acoust. Soc. Am., 97. • Acknowledgements: • This work was supported by grant DC 02012 from the National Institutes of Health. Target (delayed) + masker The target tone pips are delayed relative to the masker tone pips, yielding both onset and offset delays. The delay required to increased d’ from 0.5 to 1 is estimated. masker alone small p masker alone large p 5000 Frequency (Hz) Frequency (Hz) 0 240 target Time (ms) Figure 4: Example of target delay 200 0 0 240 240 5 target frequency Time (ms) 4 Figure 2: Increase in p leads to lower d’ 3 Sound source segregation is determined by a variety of aspects of the environment, including harmonicity of the target (if complex), onset/offset of the target relative to competing organizations, relatively stable frequency trajectory over time, etc. In the present experiments the effects of (a) Target frequency uncertainty, (b) Relative onset/offset delay (target pips vs. masker pips) (c) joint effects of masker density (p) and onset/offset delay are studied. Threshold delay in ms (d’ = 1) 2 1 0 445 1000 2245 4490 220 Low MidLow Mid MidHigh High Ran Target Frequency (Hz) Figure 5: Threshold delay as a function of frequency and frequency uncertainty