Denoising audio examples

Below, examples named “Plain Cosparse” or “Social Cosparse” are obtained with our general audio reconstruction framework while examples named “BT” are the result of the well known Block-Thresholding* method.

*Yu, G., Mallat, S., & Bacry, E. (2008). Audio denoising by time-frequency block thresholding. IEEE Transactions on Signal processing, 56(5), 1830-1839.
 

Pop music

Input SNR Noisy BT Plain Cosparse Social Cosparse Clean
5dB
10dB
15dB
20dB

Jazz music

Input SNR Noisy BT Plain Cosparse Social Cosparse Clean
5dB
10dB
15dB
20dB

Chamber music

Input SNR Noisy BT Plain Cosparse Social Cosparse Clean
5dB
10dB
15dB
20dB

Speech

Input SNR Noisy BT Plain Cosparse Social Cosparse Clean
5dB
10dB
15dB
20dB

 

Comments are closed.