3 March 2023 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
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.
23 June 2021 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
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.
22 December 2016 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
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.
1 April 2017 in The Journal of the Acoustical Society of America
The efficient measurement of the threshold and slope of the psychometric function (PF) is an important objective in psychoacoustics. This paper proposes a procedure that combines a Bayesian estimate of the PF with either a look one-ahead or a look two-ahead method of selecting the next stimulus presentation. The procedure differs from previously proposed algorithms in two respects: (i) it does not require the range of possible PF parameters to be specified in advance and (ii) the sequence of probe signal-to-noise ratios optimizes the threshold and slope estimates at a performance level, ϕ, that can be chosen by the experimenter. Simulation results show that the proposed procedure is robust and that the estimates of both threshold and slope have a consistently low bias. Over a wide range of listener PF parameters, the root-mean-square errors after 50 trials were ∼1.2 dB in threshold and 0.14 in log-slope. It was found that the performance differences between the look one-ahead and look two-ahead methods were negligible and that an entropy-based criterion for selecting the next stimulus was preferred to a variance-based criterion.