Table 6. Comparison against non-deep learning methods.

Test Scores are averaged over five noise sources at all SNRs.

	STOI (in %)				PESQ				SDR
Algorithm	-5dB	0dB	5dB	Avg	-5dB	0dB	5dB	Avg	-5dB	0dB	5dB	Avg
Noisy (UnP)	58.02	68.87	80.05	68.98	1.43	1.73	2.01	1.72	-4.73	0.13	5.07	0.13
DNN-IRM	73.94	80.06	85.39	79.79	1.83	2.23	2.61	2.22	3.84	6.40	8.50	6.25
LSTM-IRM	77.39	83.15	87.89	82.81	2.02	2.36	2.72	2.37	4.00	6.61	9.35	6.65
LLMSE	59.21	70.22	81.80	70.41	1.49	1.75	2.22	1.82	-3.43	0.18	5.32	0.69
OM-LSA	59.49	71.50	82.14	71.04	1.52	1.81	2.20	1.84	-3.88	0.21	5.51	0.61
Proposed-M	82.46	87.20	91.84	87.16	2.24	2.62	2.92	2.59	4.19	7.17	10.3	7.20
Proposed-UM	79.06	84.08	89.48	84.20	2.09	2.40	2.79	2.43	4.09	6.72	10.0	6.97
Proposed-Avg	80.76	85.64	90.66	85.68	2.17	2.51	2.85	2.51	4.14	6.95	10.2	7.09