Mixture of Inference Networks for VAE-based Audio-visual Speech Enhancement

by Mostafa Sadeghi, Xavier Alameda-Pineda IEEE TSP, 2021 [paper] [arXiv] Abstract. In this paper, we are interested in unsupervised (unknown noise) speech enhancement, where the probability distribution of clean speech spectrogram is simulated via a latent variable generative model, also called the decoder. Recently, variational autoencoders (VAEs) have gained much popularity…

Continue reading

NAOLab toolbox

NAOLab is a toolbox developed within the EARS EU project. NAOLab enables easy robot programming of the NAO humanoid robot (v5), using external modules developed in C, C++, Python, or Matlab. Please visit the NAOLab page for more details and for software download.

Continue reading

Mixcam Software

The Mixcam Software has been developed over the two past years, by the engineers of the Perception team : Michel Amat Pierre Arquier Quentin Pelorson The software is dedicated to the Mixcam Laboratory. The software provides an easy-to-use interface to researchers for 3D reconstruction algorithms/visualization and multiple cameras  data acquisitions. Functionalities…

Continue reading

Spectral Matching

SpecMatch is an open-source software (OSS) package that performs graph matching using Laplacian embedding followed by point registration. Software developed by Diana Mateus, Avinash Sharma, David Knossow and Radu Horaud. The code can be downloaded from this page: http://open-specmatch.gforge.inria.fr/ Publications: Avinash Sharma, Radu Horaud, Diana Mateus. 3D Shape Registration Using Spectral Graph…

Continue reading

The gtde MATLAB toolbox

The gtde toolbox is a set of MATLAB functions for localizing sound sources from time delay estimates. From the sound recorded on the microphones of any non-coplanar arbitrarily-shaped microphone array, the toolbox can be used to robustly recover the position of the sound source and the time delay estimates associated…

Continue reading