TABLE II.

Objective scores based on acoustic signals for direct-sound (DS) algorithm performance. Direct-sound target speech was used as the reference.

Input TIR	Unprocessed STOI (%)	Processed STOI (%)	Benefit STOI	Unprocessed PESQ	Processed PESQ	Benefit PESQ
−6 dB	45.56	78.69	33.13	1.54	2.30	0.76
−3 dB	50.83	81.43	30.60	1.60	2.41	0.81
0 dB	56.09	83.56	27.47	1.66	2.50	0.84
3 dB	61.00	85.15	24.15	1.76	2.58	0.82
6 dB	65.11	86.27	21.16	1.88	2.67	0.79
Mean	55.72	83.02	27.30	1.69	2.49	0.80