Denoiser samples (current model architecture based on FullSubNet+ )
This webpage presents some results of the trained models on singing voice.
Models 1 and 2 have the same architecture, the only difference being the Signal-to-Noise Ratio (SNR) range used for dynamic mixing of the training dataset.
SNR range for model 1 = [-5,20] dB
SNR range for model 2 = [10,40] dB
SNR range for model 3 = [-5,40] dB
Stationary noise: air conditioning
Orignal
SNR = 30 dB
SNR = 20 dB
SNR = 10 dB
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 1 (original)
Enhanced by model 1 (SNR = 30 dB)
Enhanced by model 1 (SNR = 20 dB)
Enhanced by model 1 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 2 (original)
Enhanced by model 2 (SNR = 30 dB)
Enhanced by model 2 (SNR = 20 dB)
Enhanced by model 2 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 3 (original)
Enhanced by model 3 (SNR = 30 dB)
Enhanced by model 3 (SNR = 20 dB)
Enhanced by model 3 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Stationary noise: fridge
Orignal
SNR = 30 dB
SNR = 20 dB
SNR = 10 dB
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 1 (original)
Enhanced by model 1 (SNR = 30 dB)
Enhanced by model 1 (SNR = 20 dB)
Enhanced by model 1 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 2 (original)
Enhanced by model 2 (SNR = 30 dB)
Enhanced by model 2 (SNR = 20 dB)
Enhanced by model 2 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 3 (original)
Enhanced by model 3 (SNR = 30 dB)
Enhanced by model 3 (SNR = 20 dB)
Enhanced by model 3 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Non-stationary noise: cat
Orignal
SNR = 30 dB
SNR = 20 dB
SNR = 10 dB
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 1 (original)
Enhanced by model 1 (SNR = 30 dB)
Enhanced by model 1 (SNR = 20 dB)
Enhanced by model 1 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 2 (original)
Enhanced by model 2 (SNR = 30 dB)
Enhanced by model 2 (SNR = 20 dB)
Enhanced by model 2 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 3 (original)
Enhanced by model 3 (SNR = 30 dB)
Enhanced by model 3 (SNR = 20 dB)
Enhanced by model 3 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Non-stationary noise: traffic
Orignal
SNR = 30 dB
SNR = 20 dB
SNR = 10 dB
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 1 (original)
Enhanced by model 1 (SNR = 30 dB)
Enhanced by model 1 (SNR = 20 dB)
Enhanced by model 1 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 2 (original)
Enhanced by model 2 (SNR = 30 dB)
Enhanced by model 2 (SNR = 20 dB)
Enhanced by model 2 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Enhanced by model 3 (original)
Enhanced by model 3 (SNR = 30 dB)
Enhanced by model 3 (SNR = 20 dB)
Enhanced by model 3 (SNR = 10 dB)
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.
Your browser does not support the audio element.