LIUTKUS , Antoine

This document is my thesis for getting the habilitation à diriger des recherches, which is the french diploma that is required to fully supervise Ph.D. students. It summarizes the research I did in the last 15 years and also provides the short­term research directions and applications I want to investigate. Regarding my past research, I first...

The increasingly widespread adoption of large language models has highlighted the need for improving their explainability. We present context length probing, a novel explanation technique for causal language models, based on tracking the predictions of a model as a function of the length of available context, and allowing to assign differential...

A python package to evaluate audio source separation results.

Source separation aims at decomposing a vector into additive components. This is often done by first estimating source parameters before feeding them into a filtering method, often based on ratios of covariances. The whole pipeline is traditionally rooted in some probabilistic framework providing both the likelihood for parameter estimation and...

We propose a new blind source separation algorithm based on mixtures of alpha-stable distributions. Complex symmetric alpha-stable distributions have been recently showed to better model audio signals in the time-frequency domain than classical Gaussian distributions thanks to their larger dynamic range. However, inference of these models is...

Audio source separation comes with the need to devise mul-tichannel filters that can exploit priors about the target signals. In that context, experience shows that modeling magnitude spectra is effective. However, devising a probabilistic model on complex spectral data with a prior on magnitudes is non trivial, because it should both reflect...

This paper reports the organization and results for the 2018 community-based Signal Separation Evaluation Campaign (SiSEC 2018). This year's edition was focused on audio and pursued the effort towards scaling up and making it easier to prototype audio separation software in an era of machine-learning based systems. For this purpose, we prepared...

This chapter presents a multichannel audio source separation framework where deep neural networks (DNNs) are used to model the source spectra and combined with the classical multichannel Gaussian model to exploit the spatial information. The parameters are estimated in an iterative expectation-maximization (EM) fashion and used to derive a...

The movement of animals is a central component of their behavioural strategies. Statistical tools for movement data analysis, however, have long been limited, and in particular, unable to account for past movement information except in a very simplified way. In this work, we propose MoveFormer, a new step-based model of movement capable of...

Music source separation is the task of decomposing music into its constitutive components, e.g., yielding separated stems for the vocals, bass, and drums. Such a separation has many applications ranging from rearranging/repurposing the stems (remixing, repanning, upmixing) to full extraction (karaoke, sample creation, audio restoration). Music...

Live concert recordings consist in long multitrack audio samples with significant interferences between channels. For audio engineering purposes, it is desirable to attenuate those interferences. Recently, we proposed an algorithm to this end based on Non-negative Matrix Factorization, that iteratively estimate the clean power spectral...

Separation of existing audio into remixable elements is very useful to repurpose music audio. Applications include upmixing video soundtracks to surround sound (e.g. home theater 5.1 systems), facilitating music transcriptions, allowing better mashups and remixes for disk jockeys, and rebalancing sound levels on multiple instruments or voices...

Protein is biology workhorse. Since the recent break-through of novel folding methods, the amount of available structural data is increasing, closing the gap between data-driven sequence-based and structure-based methods. In this work, we focus on the inverse folding problem that consists in predicting an amino-acid primary sequence from...

We propose a semi-supervised multichannel speech enhancement system based on a probabilistic model which assumes that both speech and noise follow the heavy-tailed multi-variate complex Cauchy distribution. As we advocate, this allows handling strong and adverse noisy conditions. Consequently, the model is parameterized by the source magnitude...

By building upon the recent theory that estab- lished the connection between implicit generative modeling (IGM) and optimal transport, in this study, we propose a novel parameter-free algo- rithm for learning the underlying distributions of complicated datasets and sampling from them. The proposed algorithm is based on a functional optimization...

his paper focuses on single-channel semi-supervised speech en-hancement. We learn a speaker-independent deep generative speechmodel using the framework of variational autoencoders. The noisemodel remains unsupervised because we do not assume prior knowl-edge of the noisy recording environment. In this context, our con-tribution is to...

Many people listen to recorded music as part of their everyday lives, for example from radio or TV programmes, CDs, downloads or increasingly from online streaming services. Sometimes we might want to remix the balance within the music, perhaps to make the vocals louder or to suppress an unwanted sound, or we might want to upmix a 2-channel...

The sigsep musdb18 data set consists of a total of 150 full-track songs of different styles and includes both the stereo mixtures and the original sources, divided between a training subset and a test subset.Its purpose is to serve as a reference database for the design and the evaluation of source separation algorithms. The objective of such...

Recent advances in Transformer models allow for unprecedented sequence lengths, due to linear space and time complexity. In the meantime, relative positional encoding (RPE) was proposed as beneficial for classical Transformers and consists in exploiting lags instead of absolute positions for inference. Still, RPE is not available for the recent...

In this study, we propose a novel probabilistic model for separating clean speech signals from noisy mixtures by decomposing the mixture spectrograms into a structured speech part and a more flexible residual part. The main novelty in our model is that it uses a family of heavy-tailed distributions, so called the α-stable distributions, for...

The Signal Separation Evaluation Campaign (SiSEC) is a large-scale regular event aimed at evaluating current progress in source separation through a systematic and reproducible comparison of the participants' algorithms, providing the source separation community with an invaluable glimpse of recent achievements and open challenges. This paper...

This paper introduces a new method for multichannel speech enhancement based on a versatile modeling of the residual noise spec-trogram. Such a model has already been presented before in the single channel case where the noise component is assumed to follow an alpha-stable distribution for each time-frequency bin, whereas the speech...

Popular music is often composed of an accompaniment and a lead component, the latter typically consisting of vocals. Filtering such mixtures to extract one or both components has many applications, such as automatic karaoke and remixing. This particular case of source separation yields very specific challenges and opportunities, including the...

Le problème de reconstruction de phase, i.e., la reconstruction d'un signal complexe à partir de mesures d'amplitudes, est un problème bien connu de la littérature. Adoptant un point de vue Bayésien et modélisant les phases non observées par des variables cachées, nous proposons ici une formulation généralisant différents cadres applicatifs,...