Skip to main content

Table 5 Evaluation results for wide-band \(RT_{60}\) and \(C_{50}\) (left channel only) for the models trained only with real training data. The input features describe the type of inputs and the size of the window

From: An end-to-end approach for blindly rendering a virtual sound source in an audio augmented reality environment

Model and input feature

\(\rho _{RT}\)

MSE\(_{RT}\)[s]

\(\rho _{C_{50}}\)

RMSE\(_{C_{50}}\)[dB]

Mel\(_{256}\) [13]

0.74

0.04

0.82

3.1

Mel\(_{512}\) [13]

0.67

0.03

0.80

3.3

+ Phase\(_{256}\) [21]

0.76

0.04

0.88

2.7

+ Phase\(_{512}\) [21]

0.75

0.04

0.79

3.1

+ Phase and continuity\(_{256}\) [21]

0.78

0.03

0.85

1.8

+ Phase and continuity\(_{512}\) [21]

0.79

0.03

0.88

2.4

+ IPD and ICD \(_{256}\)

0.87

0.01

0.93

1.9

+ IPD and ICD \(_{512}\)

0.85

0.01

0.84

2.9