Quantifying Informational Masking due to Masker Intelligibility in Same-talker Speech-in-speech Perception

Intelligibility of the competing speech plays a significant role in causing informational masking (IM) to the target speech during speech-in-speech perception, especially in same-talker conditions where the target and the masker share a large number of similarities in acoustics. Few studies have quantitatively measured IM as a function of intelligibility of competing speech. Evidence shows that voiced segments are robust cues for speech intelligibility. In this study, the contribution of masker intelligibility to IM was studied by adjusting the voice-to-noise ratio (VNR) on voiced segments of the competing speech, while maintaining energetic masking (EM) at different target-to-masker ratios. Although model estimations suggested that the intelligibility due to EM converged when VNR<0 dB, listener performance showed that more release from IM was received with a further decrease in VNR. It was projected that masker intelligibility could lead to target intelligibility decreased by 50%.