Speech Denoising by Accumulating Per-Frequency Modeling Fluctuations
This addresses audio denoising for applications like speech processing, but it is incremental as it builds on existing unsupervised and time-frequency domain techniques.
The paper tackles the problem of audio denoising by proposing an unsupervised method that trains a deep neural network on a single noisy audio clip to disentangle clean signals, achieving favorable performance compared to existing literature methods.
We present a method for audio denoising that combines processing done in both the time domain and the time-frequency domain. Given a noisy audio clip, the method trains a deep neural network to fit this signal. Since the fitting is only partly successful and is able to better capture the underlying clean signal than the noise, the output of the network helps to disentangle the clean audio from the rest of the signal. This is done by accumulating a fitting score per time-frequency bin and applying the time-frequency domain filtering based on the obtained scores. The method is completely unsupervised and only trains on the specific audio clip that is being denoised. Our experiments demonstrate favorable performance in comparison to the literature methods. Our code and samples are available at github.com/mosheman5/DNP and as supplementary. Index Terms: Audio denoising; Unsupervised learning