. 2015 Dec 17;19:2331216515617917. doi: 10.1177/2331216515617917

Table A1.

SCNR Parameter Settings Tested in the Instrumental (iSNR) Evaluation.

	SCNR 1	SCNR 2	SCNR 3	SCNR 4
G_min	−9 dB	−9 dB	−12.5 dB	−17 dB
$α^{const} (q)$	0.5 if q $\in {0, \dots, 2}$	0.2 if q $\in {0, \dots, 2}$	0.2 if q $\in {0, \dots, 2}$	0.2 if q $\in {0, \dots, 2}$
	0.7 if q $\in {3, \dots, 19}$	0.4 if q $\in {3, \dots, 19}$	0.4 if q $\in {3, \dots, 19}$	0.4 if q $\in {3, \dots, 19}$
	0.97 if q $\in {20, \dots, 256}$	0.92 if q $\in {20, \dots, 256}$	0.92 if q $\in {20, \dots, 256}$	0.92 if q $\in {20, \dots, 256}$
$α_{pitch}$	0.2	0.15	0.15	0.15

Note. G_min denotes the lower limit of the gain function, $α^{const} (q)$ the smoothing coefficient, dependent on the cepstral index q and $α_{pitch}$ the smoothing coefficient, which is applied to the cepstral component where the fundamental frequency of the speech is detected.