A multichannel learning-based approach for sound source separation in reverberant environments

EURASIP Journal on Audio, Speech, and Music Processing

Table 7 Performance improvement for UCAs with different number of microphones when applied in rooms with different reverberation times

Channel	Model	ΔSI-SNR (dB)					ΔPESQ					ΔSTOI
Channel	Model	0.16 s	0.36 s	0.61 s	0.9 s	Avg.	0.16 s	0.36 s	0.61 s	0.9 s	Avg	Avg.
2-ch	BF-net + U-net + DM	6.92	8.25	8.75	8.54	8.10	0.24	0.26	0.25	0.19	0.24	0.10
3-ch		8.41	9.72	10.84	10.63	9.88	0.42	0.39	0.33	0.28	0.36	0.14
4-ch		9.55	11.06	11.69	11.7	11.01	0.49	0.50	0.37	0.34	0.43	0.16
6-ch		12.33	13.78	15.02	14.98	13.99	1.10	0.90	0.70	0.59	0.83	0.22
Score of unprocessed signals		− 5.67	− 11.45	− 16.17	− 18.27	− 12.73	1.80	1.59	1.43	1.36	1.55	0.59