arXiv subjects

Efthymios Tzinis

Publications and source records attributed to Efthymios Tzinis.

29 records · Page 2Linked to original sources

arXiv2020

Sudo rm -rf: Efficient Networks for Universal Audio Source Separation

In this paper, we present an efficient neural network for end-to-end general purpose audio source separation. Specifically, the backbone structure of this convolutional network is the SUccessive DOwnsampling and Resampling of Multi-Resolution Features (SuDoRMRF) as well as their aggregation which is performed through simple one-dimensional convolutions. In this way, we are able to obtain high quality audio source separation with limited number of floating point operations, memory requirements, number of parameters and latency. Our experiments on both speech and environmental sound separation datasets show that SuDoRMRF performs comparably and even surpasses various state-of-the-art approaches with significantly higher computational resource requirements.

eess.AS↗

arXiv2020

Asteroid: the PyTorch-based audio source separation toolkit for researchers

This paper describes Asteroid, the PyTorch-based audio source separation toolkit for researchers. Inspired by the most successful neural source separation systems, it provides all neural building blocks required to build such a system. To improve reproducibility, Kaldi-style recipes on common audio source separation datasets are also provided. This paper describes the software architecture of Asteroid and its most important features. By showing experimental results obtained with Asteroid's recipes, we show that our implementations are at least on par with most results reported in reference papers. The toolkit is publicly available at https://github.com/mpariente/asteroid .

eess.AS↗

arXiv2019

Improving Universal Sound Separation Using Sound Classification

Deep learning approaches have recently achieved impressive performance on both audio source separation and sound classification. Most audio source separation approaches focus only on separating sources belonging to a restricted domain of source classes, such as speech and music. However, recent work has demonstrated the possibility of "universal sound separation", which aims to separate acoustic sources from an open domain, regardless of their class. In this paper, we utilize the semantic information learned by sound classifier networks trained on a vast amount of diverse sounds to improve universal sound separation. In particular, we show that semantic embeddings extracted from a sound classifier can be used to condition a separation network, providing it with useful additional information. This approach is especially useful in an iterative setup, where source estimates from an initial separation stage and their corresponding classifier-derived embeddings are fed to a second separation network. By performing a thorough hyperparameter search consisting of over a thousand experiments, we find that classifier embeddings from clean sources provide nearly one dB of SNR gain, and our best iterative models achieve a significant fraction of this oracle performance, establishing a new state-of-the-art for universal sound separation.

cs.SD↗

arXiv2019

Pattern Search Multidimensional Scaling

We present a novel view of nonlinear manifold learning using derivative-free optimization techniques. Specifically, we propose an extension of the classical multi-dimensional scaling (MDS) method, where instead of performing gradient descent, we sample and evaluate possible "moves" in a sphere of fixed radius for each point in the embedded space. A fixed-point convergence guarantee can be shown by formulating the proposed algorithm as an instance of General Pattern Search (GPS) framework. Evaluation on both clean and noisy synthetic datasets shows that pattern search MDS can accurately infer the intrinsic geometry of manifolds embedded in high-dimensional spaces. Additionally, experiments on real data, even under noisy conditions, demonstrate that the proposed pattern search MDS yields state-of-the-art results.

cs.LG↗

arXiv2019

End-to-end Non-Negative Autoencoders for Sound Source Separation

Discriminative models for source separation have recently been shown to produce impressive results. However, when operating on sources outside of the training set, these models can not perform as well and are cumbersome to update. Classical methods like Non-negative Matrix Factorization (NMF) provide modular approaches to source separation that can be easily updated to adapt to new mixture scenarios. In this paper, we generalize NMF to develop end-to-end non-negative auto-encoders and demonstrate how they can be used for source separation. Our experiments indicate that these models deliver comparable separation performance to discriminative approaches, while retaining the modularity of NMF and the modeling flexibility of neural networks.

cs.SD↗

arXiv2019

Two-Step Sound Source Separation: Training on Learned Latent Targets

In this paper, we propose a two-step training procedure for source separation via a deep neural network. In the first step we learn a transform (and it's inverse) to a latent space where masking-based separation performance using oracles is optimal. For the second step, we train a separation module that operates on the previously learned space. In order to do so, we also make use of a scale-invariant signal to distortion ratio (SI-SDR) loss function that works in the latent space, and we prove that it lower-bounds the SI-SDR in the time domain. We run various sound separation experiments that show how this approach can obtain better performance as compared to systems that learn the transform and the separation module jointly. The proposed methodology is general enough to be applicable to a large class of neural network end-to-end separation systems.

cs.LG↗

arXiv2019

Continual Learning of New Sound Classes using Generative Replay

Continual learning consists in incrementally training a model on a sequence of datasets and testing on the union of all datasets. In this paper, we examine continual learning for the problem of sound classification, in which we wish to refine already trained models to learn new sound classes. In practice one does not want to maintain all past training data and retrain from scratch, but naively updating a model with new data(sets) results in a degradation of already learned tasks, which is referred to as "catastrophic forgetting." We develop a generative replay procedure for generating training audio spectrogram data, in place of keeping older training datasets. We show that by incrementally refining a classifier with generative replay a generator that is 4% of the size of all previous training data matches the performance of refining the classifier keeping 20% of all previous training data. We thus conclude that we can extend a trained sound classifier to learn new classes without having to keep previously used datasets.

cs.LG↗

arXiv2019

A Style Transfer Approach to Source Separation

Training neural networks for source separation involves presenting a mixture recording at the input of the network and updating network parameters in order to produce an output that resembles the clean source. Consequently, supervised source separation depends on the availability of paired mixture-clean training examples. In this paper, we interpret source separation as a style transfer problem. We present a variational auto-encoder network that exploits the commonality across the domain of mixtures and the domain of clean sounds and learns a shared latent representation across the two domains. Using these cycle-consistent variational auto-encoders, we learn a mapping from the mixture domain to the domain of clean sounds and perform source separation without explicitly supervising with paired training examples.

cs.SD↗

arXiv2019

Bootstrapped Coordinate Search for Multidimensional Scaling

In this work, a unified framework for gradient-free Multidimensional Scaling (MDS) based on Coordinate Search (CS) is proposed. This family of algorithms is an instance of General Pattern Search (GPS) methods which avoid the explicit computation of derivatives but instead evaluate the objective function while searching on coordinate steps of the embedding space. The backbone element of CSMDS framework is the corresponding probability matrix that correspond to how likely is each corresponding coordinate to be evaluated. We propose a Bootstrapped instance of CSMDS (BS CSMDS) which enhances the probability of the direction that decreases the most the objective function while also reducing the corresponding probability of all the other coordinates. BS CSMDS manages to avoid unnecessary function evaluations and result to significant speedup over other CSMDS alternatives while also obtaining the same error rate. Experiments on both synthetic and real data reveal that BS CSMDS performs consistently better than other CSMDS alternatives under various experimental setups.

cs.LG↗

arXiv2018

Unsupervised Deep Clustering for Source Separation: Direct Learning from Mixtures using Spatial Information

We present a monophonic source separation system that is trained by only observing mixtures with no ground truth separation information. We use a deep clustering approach which trains on multi-channel mixtures and learns to project spectrogram bins to source clusters that correlate with various spatial features. We show that using such a training process we can obtain separation performance that is as good as making use of ground truth separation information. Once trained, this system is capable of performing sound separation on monophonic inputs, despite having learned how to do so using multi-channel recordings.

cs.LG↗

arXiv2018

Integrating Recurrence Dynamics for Speech Emotion Recognition

We investigate the performance of features that can capture nonlinear recurrence dynamics embedded in the speech signal for the task of Speech Emotion Recognition (SER). Reconstruction of the phase space of each speech frame and the computation of its respective Recurrence Plot (RP) reveals complex structures which can be measured by performing Recurrence Quantification Analysis (RQA). These measures are aggregated by using statistical functionals over segment and utterance periods. We report SER results for the proposed feature set on three databases using different classification methods. When fusing the proposed features with traditional feature sets, we show an improvement in unweighted accuracy of up to 5.7% and 10.7% on Speaker-Dependent (SD) and Speaker-Independent (SI) SER tasks, respectively, over the baseline. Following a segment-based approach we demonstrate state-of-the-art performance on IEMOCAP using a Bidirectional Recurrent Neural Network.

cs.SD↗