
How to use KLdivergence to construct conjugate priors, with welldefined noninformative limits, for the multivariate Gaussian
The Wishart distribution is the standard conjugate prior for the precisi...
read it

The Phonexia VoxCeleb Speaker Recognition Challenge 2021 System Description
We describe the Phonexia submission for the VoxCeleb Speaker Recognition...
read it

Out of a hundred trials, how many errors does your speaker verifier make?
Out of a hundred trials, how many errors does your speaker verifier make...
read it

A Speaker Verification Backend with Robust Performance across Conditions
In this paper, we address the problem of speaker verification in conditi...
read it

Probabilistic embeddings for speaker diarization
Speaker embeddings (xvectors) extracted from very short segments of spe...
read it

LargeScale Speaker Diarization of Radio Broadcast Archives
This paper describes our initial efforts to build a largescale speaker ...
read it

Fast variational Bayes for heavytailed PLDA applied to ivectors and xvectors
The standard stateoftheart backend for textindependent speaker recog...
read it

Gaussian metaembeddings for efficient scoring of a heavytailed PLDA model
Embeddings in machine learning are lowdimensional representations of co...
read it

Languagedepedent IVectors for LRE15
A standard recipe for spoken language recognition is to apply a Gaussian...
read it

A Generative Model for Score Normalization in Speaker Recognition
We propose a theoretical framework for thinking about score normalizatio...
read it

Note on the equivalence of hierarchical variational models and auxiliary deep generative models
This note compares two recently published machine learning methods for c...
read it

VB calibration to improve the interface between phone recognizer and ivector extractor
The EM training algorithm of the classical ivector extractor is often i...
read it

Constrained speaker linking
In this paper we study speaker linking (a.k.a. partitioning) given const...
read it

Bayesian calibration for forensic evidence reporting
We introduce a Bayesian solution for the problem in forensic speaker rec...
read it

A comparison of linear and nonlinear calibrations for speaker recognition
In recent work on both generative and discriminative score to loglikeli...
read it

The EM algorithm and the Laplace Approximation
The Laplace approximation calls for the computation of second derivative...
read it

Likelihoodratio calibration using priorweighted proper scoring rules
Priorweighted logistic regression has become a standard tool for calibr...
read it

Generative, Fully Bayesian, Gaussian, Openset Pattern Classifier
This report works out the details of a closedform, fully Bayesian, mult...
read it

The BOSARIS Toolkit: Theory, Algorithms and Code for Surviving the New DCF
The change of two orders of magnitude in the 'new DCF' of NIST's SRE'10,...
read it

The PAV algorithm optimizes binary proper scoring rules
There has been much recent interest in application of the pooladjacent...
read it