Fig. 5.

Validity of web-based testing for demonstrating the co-modulation masking release (CMR) effect. A CMR was measured by assessing the SNR thresholds for detecting 4-kHz tones in modulated 1-ERB-wide on-band noise for varying configurations of 1-ERB-wide noise flankers (2-ERB gap between on-band noise and flankers). The flankers were either absent (REF), modulated in a correlated manner with the on-band noise (CORR), or anticorrelated manner with the on-band noise (ACORR). The change in tone thresholds across conditions is quantified as the CMR effect. (B) Psychometric curves for tone detection in different CMR conditions was measured from $N=203$ subjects, yielding a clear separation between conditions. CMR for CORR-REF was about 4 dB, whereas CMR for CORR - ACORR was about 17 dB consistent with lab-based measurements with identical stimuli. Horizontal error bars indicate 95% confidence intervals for the tone threshold in each condition