Skip to the content.

Irregular Circular Array Results


In this page we show results corresponding to reproduction of a single source when using the deep learning-based model presented in [1], denoted as CNN, the Model-based rendering technique presented in [2] denoted as MR, the pressure-matching approach [3] denoted as PM and the adaptive wavefield synthesis technique[4], denoted as AWFS. We also present results related to an adaptive version of MR denoted AMR. We consider a regular loudspeaker array of 64 elements and randomly remove 16,32, and 48 loudspeakers, ending with a 48,32,16 element array, respectively

Training/test setup

In the image we show the setup used to generate the training and test data. real soundfield

Ground truth

real soundfield

48 loudspeaker setup

real soundfield

32 loudspeaker setup

real soundfield

16 loudspeaker setup

real soundfield


[1] Comanducci, L. Antonacci, F., & Sarti. A., Synthesis of Soundfields through Irregular Loudspeaker Arrays Based on Convolutional Neural Networks [arXiv preprint].

[2] Bianchi, L., Antonacci, F., Sarti, A., & Tubaro, S. (2016). Model-based acoustic rendering based on plane wave decomposition. Applied Acoustics, 104, 127-134.

[3] Nelson, P. A. (1994). Active control of acoustic fields and the reproduction of sound. Journal of Sound and Vibration, 177(4), 447-477.

[4] Gauthier, Philippe-Aubert, and Alain Berry. “Adaptive wave field synthesis with independent radiation mode control for active sound field reproduction: Theory.” The Journal of the Acoustical Society of America 119.5 (2006): 2721-2737.

Support or Contact

For any information regarding the paper or the code send us an email at!