TAKANEN2013_WBMSO - Wideband MSO model

Usage:

[output energy] = takanen2013_wbmso(ipsilateral, contralateral, fs, widthinerbs, fc, printfigs)

Input parameters:

`ipsilateral`	The ipsilateral "where" stream output from the model of the periphery
`contralateral`	The contralateral "where" stream output from the model of the periphery
`fs`	Sampling rate
`widthinerbs`	The number of adjacent ERB bands the information is gathered over
`fc`	Characteristic frequencies
`printFigs`	Boolean value that defines whether several figures illustrating the processing steps in the model are plotted or not. As default, no figures are plotted.

Output parameters:

`output`	Spatial cues for frequency bands that are summed together to form a width defined by widthinerbs
`energy`	"What" stream for the wideband MSO

Description:

The wideband MSO model simulates the ability of the human auditory system to extract localization cues based on interaural envelope time shifts. These time shifts are decoded into directional cues for the model. This is done by processing the output of the periphery model with the following steps:

Delay the contralateral signal.
Adjacent frequency bands are summed on both sides.
The average of the signal is removed to emphasize prominent peaks by applying a self-weighted moving average filter and delay to the signal and deducting this from the signal after summing over frequency bands.
Both sides are convolved with a Hanning window and a phase-locked impulse generator is applied.
The contralateral side is convolved and the ipsilateral side is limited and normalized.
Coincidence detection between the ipsilateral and contralateral signals.
Weighted and self-weighted moving average filters are applied to the outputs of the coincidence detection and contralateral signal, respectively, and the output is limited.

References:

V. Pulkki and T. Hirvonen. Functional count-comparison model for binaural decoding. Acta Acustica united with Acustica, 95(5):883 -- 900, Sept./Oct. 2009.