Embedded Files
Legend:
Original Audio: Normal noisy input to the model, x
Initial Model Output: Normal model behavior f (x), i.e., ŷ
Attacked Audio: Original audio plus a perturbation convolved with the same RIR: x + r ∗ δ
Attacked Model Output: f (x + r ∗ δ)
Demucs (Denoiser)
Demucs (Denoiser)
Original Audio
Initial Model Output
adversarial_speech_epoch_0.wavadversarial_out_epoch_0.wavAttacked Audio
Attacked Model Output
adversarial_speech_final.wavadversarial_out_final.wavMP-SENet
MP-SENet
Original Audio
Initial Model Output
adversarial_speech_epoch_0.wavadversarial_out_epoch_0.wavAttacked Audio
Attacked Model Output
adversarial_speech_final.wavadversarial_out_final.wavFRCRN
FRCRN
Original Audio
Initial Model Output
adversarial_speech_epoch_0.wavadversarial_out_epoch_0.wavAttacked Audio
Attacked Model Output
adversarial_speech_final.wavadversarial_out_final.wavPage updated
Google Sites
Report abuse