Usage:

[processed, mpar] = eurich2022(mRef, mTest,mpar)

Input parameters:

Output parameters:

Description:

eurich2022 is a binaural detection model based on two-hemisphere code. It involves interaural incoherence interference. For binaural reference and test signals, it computes the complex correlation coefficient (gamma) representing the two-hemisphere encoding found in mammals' medial superior olives (MSO). To account for degraded binaural detection for maskers with frequency-dependent interaural coherence, an across-frequency interference mechanisms of interaural incoherence is introduced. Additionally, the model computes the DC power to account for energetic unmasking. eurich2022_processing computes these features based on a binaural signal. Then, eurich2022_decision derives the model's sensitivity based on these computed features.

mpar is an optional structure with the following parameters:

• fs: sampling frequency of model
• Filters_per_ERB_aud: spacing of peripheral filter central frequencies in ERB
• GT_bandwidth_factor: factor of gammatone filter bandwidth relative to "standard" (i.e. ERB = 79 Hz @500Hz)
• GT_lowest_center_frequency: lower bound of gammatone filterbank
• GT_highest_center_frequency: higher bound of gammatone filterbank
• GT_fix_center_frequency: fixed center frequency of one of the gammatone filterbank's filters
• GT_filterorder: filter order of gammatone filterbank
• interference_sigma: std of the Gaussian window used for the across-frequency incoherence interference
• iKernelThresh: threshold above which a value of the Gaussian filter window is used
• rho_max upper limit of encoded interaural coherence (i.e. internal noise)
• monaural_internal_noise_sigma: std of the Gaussian, level-dependent internal noise of the monaural feature
• binaural_internal_noise_sigma: std of the Gaussian internal noise of the binaural feature
• target_channel: index of the channel to be selected for processing in eurich2022_decision