[benefit weighted_SNR weighted_bmld] = jelfs2011(target,interferer,fs)
|target||Binaural target impulse response (or stimulus)|
|interfererer||Binaural interferer impulse response (or stimulus) Multiple interfering impulse responses MUST be concatenated, not added.|
|benefit||spatial release from masking (SRM)in dB|
|weighted_SNR||component of SRM due to better-ear listening (dB)|
|weighted_bmld||component of SRM due to binaural unmasking (dB)|
jelfs2011(target,interferer,fs) computes the increase in speech intelligibility of the target when the target and interferer are spatially separated. They are preferably represented by their impulse responses, but can be represented by noise recordings of equivalent spectral shape emitted from the same source locations (using the same noise duration for target and interferer). The impulse responses are assumed to be sampled at a sampling frequency of fs Hz. If the modelled sources differ in spectral shape, this can be simulated by pre-filtering the impulse responses.
[benefit, weighted_SNR, weighted_bmld]=jelfs2011(...) additionally returns the benefit from the SII weighted SNR and the SII weighted BMLD.
If target or interferer are cell-arrays, the HRTF data will be loaded. The first argument in the cell-array is the azimuth angle, and the second parameter is the database type. The elevation is set to zero. function.
The following code will load HRIRs from the 'kemar' database and compute the binaural speech intelligibility advantage for a target at 0 degrees and interferers at 300 and 90 degrees:
This code produces the following output:
ans = 1.0773
M. Lavandier, S. Jelfs, J. Culling, A. Watkins, A. Raimond, and S. Makin. Binaural prediction of speech intelligibility in reverberant rooms with multiple noise sources. J. Acoust. Soc. Am., 131(1):218--231, 2012.