# Experiments on Open-Set Speaker Identification with Discriminatively Trained Neural Networks

## TL;DR

This study evaluates neural network classifiers for open-set speaker identification, demonstrating that multi-class neural networks outperform traditional Gaussian mixture models, especially with larger speaker populations.

## Contribution

The paper provides a comparative analysis of neural network architectures versus Gaussian mixture models in open-set speaker identification, highlighting the advantages of multi-class neural networks.

## Key findings

- Multi-class neural networks outperform GMMs for large speaker populations.
- Neural networks show improved accuracy in open-set conditions.
- Performance varies with the size of enrolled speaker sets.

## Abstract

This paper presents a study on discriminative artificial neural network classifiers in the context of open-set speaker identification. Both 2-class and multi-class architectures are tested against the conventional Gaussian mixture model based classifier on enrolled speaker sets of different sizes. The performance evaluation shows that the multi-class neural network system has superior performance for large population sizes.