Skip to main content

Table 1 Source separation results for 1 s of speech

From: Perceptually controlled doping for audio source separation

No. of sources

SDR (dB)

SDR sparse (dB)

SIR (dB)

SIR sparse (dB)

SAR (dB)

SAR sparse (dB)

2

50.4

49.9

60.4

59.8

51.5

51.2

3

14.0

15.1

18.6

19.8

16.3

17.5

4

9.5

10.7

13.5

14.7

12.2

13.4

5

5.8

6.6

9.1

9.8

9.7

10.7

6

3.5

4.5

6.5

7.3

8.2

9.3

  1. SDR, SIR, and SAR for original and sparsified signals. Mean values of 100 simulations with 1-s speech signals.