Supervised deep learning approaches to underdetermined audio source separation achieve state-of-the-art performance but require a dataset of mixtures along with their corresponding isolated source signals. Such datasets can be extremely costly to obtain for musical mixtures. This raises a need for unsupervised methods. We propose a novel unsupervised model-based deep learning approach to musical source separation. Each source is modelled with a differentiable parametric source-filter model. A neural network is trained to reconstruct the observed mixture as a sum of the sources by estimating the source models' parameters given their fundamental frequencies. At test time, soft masks are obtained from the synthesized source signals. The experimental evaluation on a vocal ensemble separation task shows that the proposed method outperforms learning-free methods based on nonnegative matrix factorization and a supervised deep learning baseline. Integrating domain knowledge in the form of source models into a data-driven method leads to high data efficiency: the proposed approach achieves good separation quality even when trained on less than three minutes of audio. This work makes powerful deep learning based separation usable in scenarios where training data with ground truth is expensive or nonexistent.

The goal of singing voice separation is to recover the vocals signal from music mixtures. State-of-the-art performance is achieved by deep neural networks trained in a supervised fashion. Since training data are scarce and music signals are extremely diverse, it remains challenging to achieve high separation quality across various recording and mixing conditions as well as music styles. In this paper, we investigate to which extent the separation can be improved when lyrics transcripts are used as additional information. To this end, we propose a joint approach to phoneme level lyrics alignment and text-informed singing voice separation. It is based on DTW-attention, a new monotonic attention mechanism including a differentiable approximation of dynamic time warping. Experimental results show that the method can align phonemes with mixed singing voice with high precision given accurate transcripts. It also achieves competitive results on challenging word level alignment test sets using less training data than state-of-the-art methods. Sequential alignment and informed separation lead to improved separation quality according to objective measures. Text information helps preserving spectral phoneme properties in the separated voice signals.

Speech separation quality can be improved by exploiting textual information. However, this usually requires text-to-speech alignment at phoneme level. Classical alignment methods are made for rather clean speech and do not work as well on corrupted speech. We propose to perform text-informed speech-music separation and phoneme alignment jointly using recurrent neural networks and the attention mechanism. We show that it leads to benefits for both tasks. In experiments, phoneme transcripts are used to improve the perceived quality of separated speech over a non-informed baseline. Moreover, our novel phoneme alignment method based on the attention mechanism achieves state-of-the-art alignment accuracy on clean and on heavily corrupted speech.

This paper proposes an online approach to the singing voice separation problem. Based on a combination of one-dimensional convolutional layers along the frequency axis and recurrent layers to enforce temporal coherency, state-of-the-art performance is achieved. The concept of using deep features in the loss function to guide training and improve the model's performance is also investigated.

This paper proposes an online single-channel speech enhancement method designed to improve the quality of speech degraded by reverberation and noise. Based on an autoregressive model for the reverberation power and on a hidden Markov model for clean speech production, a Bayesian filtering formulation of the problem is derived and online joint estimation of the acoustic parameters and mean speech, reverberation, and noise powers is obtained in mel-frequency bands. From these estimates, a real-valued spectral gain is derived and spectral enhancement is applied in the short-time Fourier transform (STFT) domain. The method yields state-of-the-art performance and greatly reduces the effects of reverberation and noise while improving speech quality and preserving speech intelligibility in challenging acoustic environments.

The reverberation of an acoustic channel can be characterised by two frequency-dependent parameters: the reverberation time and the direct-to-reverberant energy ratio. This paper presents an algorithm for blindly determining these parameters from a single-channel speech signal. The algorithm uses an extended Kalman filter to estimate the parameters together with a hidden semi-Markov model to identify intervals of speech activity.