Scott K. Isabelle

Binaural detection with reproducible narrow‐band maskers

Responses of five subjects for the binaural detection of a 500‐Hz, 300‐ms, interaurally out‐of‐phase tone masked by statistically independent samples of narrow‐band, interaurally identical noise are reported. A set of 30 noise samples were taken from a noise process with a power spectrum that is 115 Hz wide, centered at 500 Hz. The rms noise pressure was 75 dB SPL. For each subject, the responses show good self‐consistency and a strong dependence on individual noise samples. (For example intrasubject correlation coefficients for subsets of the results were typically 0.85 and always higher than 0.72). However, there are significant differences between subjects in the pattern of responses across the set of masker samples. (Intersubject correlation coefficients were lower than 0.7). Subjects’ performance values for individual noise samples are compared to the predictions of psychophysical models for binaural detection. Sample‐level predictions of models for which the decision variable is dominated by the energy in the particular noise sample have no significant correlation with the patterns of responses across noise samples for any subject. In contrast, models that are based on variability of the interaural differences have predictions that are statistically significantly correlated with most subjects’ responses (although no correlation coefficients were larger than 0.7). [Work supported by NIDCD (Grant DC00100).]

The cocktail party effect with conversational speech

Most often the cocktail party effect has been evaluated for relatively simple stimuli (e.g., isolated words or simple sentences masked by noise) in anechoic environments. Here, a stimulus situation is used which is similar to that employed by Plomp [Acustica 34, 200–211 (1976)] (i.e., continuous discourse for both target and masker) in a reverberant room. However, instead of asking subjects to make a subjective judgment of intelligibility, the ‘‘tracking procedure’’ of DeFillipo and Scott is used [J. Acoust. Soc. Am. 63, 1186–1192 (1978)] to simulate a conversation. This procedure measures the number of words per minute (WPM) successfully transmitted between the target talker and the subject. Preliminary results show increases in WPM of about 20% when the masker is separated from the target by 90 deg in the horizontal plane relative to the case when both the target and masker are at the same location in front of the subject. (The change in S/N necessary to maintain constant WPM between these conditions in...

Audition and the sense of presence in virtual environments

Historical reports from suddenly deafened adults describing a sense of detachment from a world devoid of auditory input suggest that auditory cues may be crucial for achieving a sense of presence in virtual environments [R. H. Gilkey and J. M. Weisenberger, Presence 4, 357–363 (1995)]. However, factors other than acoustic fidelity can affect the perceived quality of a virtual auditory display. For example, Simpson et al. [J. Acoust. Soc. Am. 100, 2633(A) (1996)] showed that the sense of presence in virtual auditory environments was driven by characteristics of the real listening environment in which the virtual audio was heard. Specifically, when the virtual auditory environment matches the real listening environment, the listening experience is particularly compelling. However, it was not clear whether auditory or visual experience with the real room mediated this effect. This paper describes attempts to separate the impact of auditory and visual information on presence by systematically manipulating the...

The experiences of untrained subjects listening to virtual sounds

Anecdotal reports from casual listening in auditory virtual environments suggest that the ‘‘compellingness’’ of virtual sounds depends not only on the fidelity of the synthesized environment but also on the characteristics of the real environment in which the sounds are presented (i.e., even though the sounds are presented through headphones, nonauditory characteristics of, or previous auditory experience in, the listening environment may influence the perception of the virtual environment). In particular, when the synthesized environment ‘‘matches’’ the listening environment the auditory experience can be strikingly realistic. To quantify this phenomenon, binaural recordings of ‘‘everyday’’ sounds were made through the KEMAR manikin in each of three real rooms (16 to 194  m3). A 3×3 between‐subjects design was used; each untrained subject listened to one of the three recordings over headphones while seated with eyes open in one of the three rooms (the recording might or might not match the room). The eff...

A model of auditory localization in noise using neural networks

Janko et al. [Binaural and Spatial Hearing in Real and Virtual Environments, edited by Gilkey and Anderson (Erlbaum, Hillsdale, NJ, 1996), pp. 557–570] used neural networks to describe auditory localization and found that data in quiet were insufficient to discriminate among models. To further constrain the models, localization in the presence of a spatially fixed masker is considered here. The wideband targets and maskers were filtered by head‐related transfer functions, then by a gammatone filter bank. A model for binaural interaction similar to that of Lindemann [J. Acoust. Soc. Am. 80, 1608–1622 (1986)] was used to process the filter‐bank outputs. The inhibited cross‐correlation output of the binaural processor was sampled and used as input to a neural network (e.g., 24 correlation lags at each of 13 frequency channels for 312 input nodes). The network was trained using back propagation to estimate the target location in quiet, then tested on targets in noise at several values of signal‐to‐noise ratio...