Blind Source Separation in Polyphonic Music Recordings Using Deep Neural Networks Trained via Policy Gradients

We propose a method for the blind separation of sounds of musical instruments in audio signals. We describe the individual tones via a parametric model, training a dictionary to capture the relative amplitudes of the harmonics. The model parameters are predicted via a U-Net, which is a type of deep neural network. The network is trained without ground truth information, based on the difference between the model prediction and the individual time frames of the short-time Fourier transform. Since some of the model parameters do not yield a useful backpropagation gradient, we model them stochastically and employ the policy gradient instead. To provide phase information and account for inaccuracies in the dictionary-based representation, we also let the network output a direct prediction, which we then use to resynthesize the audio signals for the individual instruments. Due to the flexibility of the neural network, inharmonicity can be incorporated seamlessly and no preprocessing of the input spectra is required. Our algorithm yields high-quality separation results with particularly low interference on a variety of different audio samples, both acoustic and synthetic, provided that the sample contains enough data for the training and that the spectral characteristics of the musical instruments are sufficiently stable to be approximated by the dictionary.

Method for the single-channel blind separation based on particle filtering

A method for single-channel blind separation of two digital signals is proposed. The method is based on particle filtering and state space model of the mixture of two signals. The effectiveness of method is verified and the advantage over the iterative Bayesian separation method is shown.

Antisparsity as a Prior in Blind Separation of Correlated Sources

<div>This letter introduces the concept of antisparse Blind Source Separation (BSS), proposing a suitable criterion based on the $\ell_\infty$ norm to explore the antisparsity feature. </div><div>The effectiveness of the criterion is theoretically demonstrated and it is also evaluated by computational simulations, which consider up to ten distinct sources with different correlation levels. Moreover, we simulated a scenario in wireless communication with binary sources, comparing our approach to the Constant Modulus algorithm. Both the theoretical and the simulation results highlight the potentiality of using antisparsity as a prior in BSS.</div>

Antisparsity as a Prior in Blind Separation of Correlated Sources

<div>This letter introduces the concept of antisparse Blind Source Separation (BSS), proposing a suitable criterion based on the $\ell_\infty$ norm to explore the antisparsity feature. </div><div>The effectiveness of the criterion is theoretically demonstrated and it is also evaluated by computational simulations, which consider up to ten distinct sources with different correlation levels. Moreover, we simulated a scenario in wireless communication with binary sources, comparing our approach to the Constant Modulus algorithm. Both the theoretical and the simulation results highlight the potentiality of using antisparsity as a prior in BSS.</div>