Back to search

Clemson University

Comparison GMM and SVM Classifier for Automatic Speaker Verification

Abstract

dc:description.abstract

The objective of this thesis is to develop automatic text-independent speaker verification systems using unconstrained telephone conversational speech. We began by performing a Gaussian Mixture Model Likelihood ratio verification task in speaker independent system as described by MIT Lincoln Lab. We next introduced a speaker dependent verification system based on speaker dependent thresholds. We then implemented the same system applying Support Vector Machine. In SVM, we used polynomial kernels and radial basis function kernels and compared the performance. For training and testing the system, we used low-level spectral features. Finally, we provided a performance assessment of these systems using the National Institute of Standards and technology (NIST) speaker recognition evaluation 2008 telephone corpora.

Degree

thesis:*
Name thesis:degree_name
Master of Science (MS)
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Electrical Engineering

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Afnan, Shamama
Contributors dc:contributor
  • Gowdy, Dr. John
  • Schalkoff, Dr. Robert
  • Baum, Dr. Carl

Subjects

dc:subject × 5

Identifiers

dc:identifier.*
Repository record dc:identifier
https://open.clemson.edu/all_theses/2228

Chain of custody

source
Harvested from
Clemson University
Base URL
open.clemson.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Afnan, Shamama. Comparison GMM and SVM Classifier for Automatic Speaker Verification. Thesis thesis, https://open.clemson.edu/all_theses/2228