Investigating auditory self-monitoring with psychoacoustic methods

D. Barac-Cikoja, K. Cole and L. Klein

Research Institute, Gallaudet University, Washington, D.C., USA

A new method for measuring an individual's sensitivity to delay in speech feedback has been developed and evaluated. It involves automated sound manipulations and testing procedures, allowing reliable and efficient estimate of an individual's delay detection threshold during auditory self-monitoring. The approach is novel in the context of self-hearing although common in other psychoacoustic testing.

The new technique consists of adaptive up-down changes of the speech feedback delay in a two-interval forced choice (2IFC) procedure, until a delay detection threshold is obtained. Specifically, subjects are asked to repeat syllable-length utterances (e.g., pa-pa-pa…) for approximately 2 seconds and to listen for a delay (an echo) in their speech feedback. In each trial, the subject produces two such utterances, one of which is delayed. The subject's task is to identify which utterance was delayed. Based on the subject's response accuracy, the size of the feedback delay is increased or decreased stepwise (making the task easier or harder, respectively), and the step size is progressively reduced across the trials (resulting in increasingly fine adjustments in the magnitude of the delay). Delay detection threshold is calculated from the delay values obtained on the last eight reversals (i.e., changes in the response accuracy, from correct to incorrect delay detection, or vice versa). Typical testing time is less than 15 minutes.

The procedure has been implemented using Capybara 320, a multi-processor sound computation engine which is controlled by the Kyma sound design language (Symbolic Sound Co). The system's modularity allows simultaneous realization of complex spectral and temporal sound alterations and dynamic control of the testing procedure.

This approach can be used for quantitative comparisons of the salience of speech feedback delay across different individuals and under different speaking/listening conditions. A study demonstrating one such application will be presented. There, sensitivity to feedback delay was investigated as a function of different signal processing strategies that may be found in a telecommunication set-up as well as in some listening devices for hard of hearing individuals.

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.

© 2005 Noldus Information Technology bv